> ## Documentation Index
> Fetch the complete documentation index at: https://lmsysorg-dsv4-1.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# GLM-5.3-Flash

> Deploy GLM-5.3-Flash with SGLang using recipes for H100, H200, B200, B300, GB200, and GB300, with MTP and multimodal serving.

export const Playground = ({config}) => {
  if (!config) {
    return <div style={{
      padding: 12,
      color: "#b91c1c"
    }}>Playground: missing <code>config</code> prop</div>;
  }
  const DIMENSIONS = ["hw", ...(config.matchDims || [{
    id: "variant"
  }, {
    id: "quant"
  }, {
    id: "strategy"
  }, {
    id: "nodes"
  }]).map(d => d.id)];
  const optionVisible = (opt, sel) => typeof opt.showWhen !== "function" || opt.showWhen(sel);
  const optionDisabled = (opt, sel) => typeof opt.disabled === "function" ? opt.disabled(sel) : !!opt.disabled;
  const visibleOptions = (spec, sel) => (spec.options || []).filter(o => optionVisible(o, sel));
  const rowVisible = (spec, sel) => (typeof spec.showWhen !== "function" || spec.showWhen(sel)) && visibleOptions(spec, sel).length > 0;
  const overlayPick = sel => {
    const picked = [];
    for (const spec of config.overlayDims || []) {
      if (!rowVisible(spec, sel)) continue;
      const opt = (spec.options || []).find(o => o.id === sel[spec.id]);
      if (opt && !optionDisabled(opt, sel)) picked.push(opt);
    }
    return picked;
  };
  const overlayPart = (sel, key) => {
    const out = [];
    for (const opt of overlayPick(sel)) {
      const add = typeof opt[key] === "function" ? opt[key](sel) : opt[key];
      if (add) out.push(...add);
    }
    return out;
  };
  const overlayCompose = (cellFlags, sel) => {
    const strip = overlayPart(sel, "stripPrefixes");
    const add = overlayPart(sel, "flags");
    if (!strip.length) return [...cellFlags || [], ...add];
    const used = new Set();
    const replacementsFor = tok => {
      const out = [];
      add.forEach((f, i) => {
        if (used.has(i) || f.split(/[\s=]/)[0] !== tok) return;
        used.add(i);
        out.push(f);
      });
      return out;
    };
    const out = [];
    for (const f of cellFlags || []) {
      const tok = f.split(/[\s=]/)[0];
      if (!strip.includes(tok)) out.push(f); else out.push(...replacementsFor(tok));
    }
    add.forEach((f, i) => {
      if (!used.has(i)) out.push(f);
    });
    return out;
  };
  const withOverlay = (cell, sel) => cell && ({
    ...cell,
    flags: overlayCompose(cell.flags, sel),
    env: [...cell.env || [], ...overlayPart(sel, "env")]
  }) || cell;
  const STORAGE_KEY = "sglang-deploy-env";
  const DEPLOYMENT_COMPONENT_ID = "deployment-configurator";
  const pgFeatures = config.playgroundFeatures || ({});
  const PD_PORTS = {
    prefill: {
      serve: 30000,
      dist: 30335
    },
    decode: {
      serve: 30100,
      dist: 30435
    }
  };
  const findCell = (cells, sel) => cells.find(c => DIMENSIONS.every(d => c.match[d] === sel[d]));
  const findMatchingCell = (cells, sel, pgEnv, pgFlags) => {
    const fixedDims = DIMENSIONS.filter(d => d !== "strategy");
    const flagsEq = (a, b) => a.length === b.length && a.every((x, i) => x === b[i]);
    const envEq = (a, b) => {
      if (a.length !== b.length) return false;
      const set = new Set(a);
      for (const x of b) if (!set.has(x)) return false;
      return true;
    };
    for (const c of cells) {
      if (fixedDims.some(d => c.match[d] !== sel[d])) continue;
      if (flagsEq(c.flags || [], pgFlags || []) && envEq(c.env || [], pgEnv || [])) {
        return c;
      }
    }
    return null;
  };
  const resolveModelName = sel => {
    const keys = [`${sel.hw}|${sel.variant}|${sel.quant}`, `${sel.variant}|${sel.quant}`, `${sel.hw}|${sel.quant}`, sel.quant, sel.hw, "default"];
    for (const k of keys) {
      const hit = config.modelNames[k];
      if (hit) return hit;
    }
    return "";
  };
  const interpolate = (text, env, modelName) => text.replace(/{{(\w+)}}/g, (_, key) => key === "MODEL_NAME" ? modelName : env[key] ?? `{{${key}}}`);
  const parseNnodes = id => {
    if (id === "single") return 1;
    const m = (/^multi-(\d+)$/).exec(id);
    return m ? parseInt(m[1], 10) : 1;
  };
  const placeholderDefaults = schema => {
    const out = {};
    for (const [k, v] of Object.entries(schema || ({}))) out[k] = v.default ?? "";
    return out;
  };
  const matchConstraint = (base, constraint) => {
    if (!constraint || typeof constraint !== "object") return false;
    const entries = Object.entries(constraint);
    if (entries.length === 0) return false;
    return entries.every(([k, vs]) => Array.isArray(vs) && vs.includes(base[k]));
  };
  const resolveRouter = (fc, sel) => {
    if (!fc || !fc.router) return null;
    const hit = (fc.routerOverrides || []).find(r => r && matchConstraint(sel, r.when));
    return hit ? {
      ...fc.router,
      ...hit
    } : fc.router;
  };
  const evaluateChip = (entry, base) => {
    if (entry === null || typeof entry !== "object") {
      return {
        value: entry,
        label: undefined,
        hidden: false,
        disabled: false,
        disableReason: ""
      };
    }
    const hidden = entry.hide ? matchConstraint(base, entry.hide) : false;
    let disabled = entry.disabled === true || entry.disable === true;
    let disableReason = typeof entry.disableReason === "function" ? entry.disableReason(base) : entry.disableReason || "";
    if (!disabled && entry.disable && typeof entry.disable === "object") {
      if (Array.isArray(entry.disable)) {
        for (const item of entry.disable) {
          const cond = item && item.when || item;
          if (matchConstraint(base, cond)) {
            disabled = true;
            if (item && item.reason) disableReason = item.reason;
            break;
          }
        }
      } else {
        disabled = matchConstraint(base, entry.disable);
      }
    }
    return {
      ...entry,
      value: entry.id !== undefined ? entry.id : entry.value,
      label: entry.label,
      hidden,
      disabled,
      disableReason
    };
  };
  const findEntry = (entries, picked) => {
    for (const e of entries || []) {
      const v = e === null || typeof e !== "object" ? e : e.id !== undefined ? e.id : e.value;
      if (v === picked) return e;
    }
    return null;
  };
  const isHidden = (entries, picked, base) => {
    const e = findEntry(entries, picked);
    if (e === null || e === undefined) return false;
    return evaluateChip(e, base).hidden;
  };
  const stripFlagsByFirstToken = (flags, prefixes) => {
    const set = new Set(prefixes);
    return flags.filter(f => !set.has(f.split(/[\s=]/)[0]));
  };
  const stripEnvByPrefix = (envList, prefixes) => {
    if (!prefixes || !prefixes.length) return envList;
    const set = new Set(prefixes);
    return envList.filter(e => !set.has(e.split("=")[0]));
  };
  const insertBeforeTail = (flags, additions) => {
    const idx = flags.findIndex(f => f.startsWith("--host"));
    const at = idx === -1 ? flags.length : idx;
    const out = flags.slice();
    out.splice(at, 0, ...additions);
    return out;
  };
  const insertAfter = (flags, afterAnyOf, additions) => {
    let idx = -1;
    for (const anchor of afterAnyOf) {
      idx = flags.findIndex(f => f.split(/[\s=]/)[0] === anchor);
      if (idx !== -1) break;
    }
    if (idx === -1) idx = flags.findIndex(f => f.startsWith("--model-path"));
    const out = flags.slice();
    out.splice(idx + 1, 0, ...additions);
    return out;
  };
  const parseIntFlag = (flags, prefix) => {
    for (const f of flags || []) {
      if (f.split(/[\s=]/)[0] !== prefix) continue;
      const rest = f.slice(prefix.length).replace(/^[\s=]+/, "");
      const n = parseInt(rest, 10);
      if (!isNaN(n)) return n;
    }
    return null;
  };
  const hasFlag = (flags, name) => (flags || []).some(f => f.split(/[\s=]/)[0] === name);
  const findFlagArg = (flags, prefix) => {
    for (const f of flags || []) {
      if (f.split(/[\s=]/)[0] !== prefix) continue;
      const rest = f.slice(prefix.length).replace(/^[\s=]+/, "");
      return rest.length ? rest : null;
    }
    return null;
  };
  const TP_HEADS = ["--tp-size", "--tp", "--tensor-parallel-size"];
  const EP_HEADS = ["--ep-size", "--ep", "--expert-parallel-size"];
  const DP_HEADS = ["--dp-size", "--dp", "--data-parallel-size"];
  const parseIntFlagAny = (flags, heads) => {
    for (const head of heads) {
      const n = parseIntFlag(flags, head);
      if (n !== null) return n;
    }
    return null;
  };
  const flagSpelling = (flags, heads, fallback) => heads.find(head => (flags || []).some(f => f.split(/[\s=]/)[0] === head)) || fallback;
  const ANCHOR_NEAR_MODEL_PATH = ["--model-path"];
  const ANCHOR_NEAR_TP = ["--tp-size", "--tp", "--model-path"];
  const ANCHOR_NEAR_DP = ["--dp-size", "--dp", "--tp-size", "--tp", "--model-path"];
  const ANCHOR_NEAR_DPATTN = ["--enable-dp-attention", "--dp-size", "--dp", "--tp-size", "--tp", "--model-path"];
  const ANCHOR_NEAR_MOE = ["--moe-a2a-backend", "--moe-runner-backend", "--enable-dp-attention", "--dp-size", "--dp", "--tp-size", "--tp", "--model-path"];
  const helpers = {
    matchConstraint,
    evaluateChip,
    findEntry,
    isHidden,
    stripFlagsByFirstToken,
    stripEnvByPrefix,
    insertBeforeTail,
    insertAfter,
    parseIntFlag,
    hasFlag,
    findFlagArg,
    TP_HEADS,
    EP_HEADS,
    DP_HEADS,
    parseIntFlagAny,
    flagSpelling,
    ANCHOR_NEAR_MODEL_PATH,
    ANCHOR_NEAR_TP,
    ANCHOR_NEAR_DP,
    ANCHOR_NEAR_DPATTN,
    ANCHOR_NEAR_MOE
  };
  const HICACHE_HEADS = ["--enable-hierarchical-cache", "--hicache-ratio", "--hicache-size", "--hicache-write-policy", "--hicache-mem-layout", "--hicache-io-backend", "--hicache-storage-backend", "--hicache-storage-prefetch-policy", "--hicache-storage-backend-extra-config"];
  const CP_ENABLE_HEADS = ["--enable-prefill-cp", "--enable-nsa-prefill-context-parallel", "--enable-dsa-prefill-context-parallel", "--enable-prefill-context-parallel"];
  const CP_MODE_HEADS = ["--nsa-prefill-cp-mode", "--dsa-prefill-cp-mode", "--prefill-cp-mode"];
  const CP_OWNED_HEADS = [...CP_ENABLE_HEADS, ...CP_MODE_HEADS, "--cp-strategy", "--attn-cp-size"];
  const CP_MODE_TO_STRATEGY = {
    "in-seq-split": "zigzag",
    "round-robin-split": "interleave"
  };
  const cpEnabledIn = flags => CP_ENABLE_HEADS.some(head => hasFlag(flags, head));
  const bakedCpStrategy = flags => findFlagArg(flags, "--cp-strategy") || CP_MODE_TO_STRATEGY[findFlagArg(flags, "--nsa-prefill-cp-mode")] || CP_MODE_TO_STRATEGY[findFlagArg(flags, "--dsa-prefill-cp-mode")] || CP_MODE_TO_STRATEGY[findFlagArg(flags, "--prefill-cp-mode")] || null;
  const AXIS_HANDLERS = {
    attention: {
      initState: () => ({
        tp: null,
        cp: null,
        cpStrategy: null,
        dpAttn: null
      }),
      deriveFromBase: (cell, fc, h) => {
        const flags = cell && cell.flags || [];
        const dpVal = h.parseIntFlagAny(flags, h.DP_HEADS);
        const hasDpAttn = h.hasFlag(flags, "--enable-dp-attention");
        let dpAttn;
        if (dpVal !== null) dpAttn = dpVal; else if (hasDpAttn) dpAttn = 1; else dpAttn = false;
        const cpSize = h.parseIntFlag(flags, "--attn-cp-size");
        return {
          tp: h.parseIntFlagAny(flags, h.TP_HEADS),
          cp: cpEnabledIn(flags) ? cpSize !== null ? cpSize : 2 : null,
          cpStrategy: bakedCpStrategy(flags),
          dpAttn
        };
      },
      apply: ({flags, env, value, fc, sel, h}) => {
        const knobEntry = id => (fc.knobs || []).find(k => k.id === id) || ({});
        const factsNow = () => ({
          ...sel || ({}),
          dpAttnOn: h.hasFlag(flags, "--enable-dp-attention"),
          cpOn: cpEnabledIn(flags),
          cpStrategy: bakedCpStrategy(flags) || "interleave",
          effTp: h.parseIntFlagAny(flags, h.TP_HEADS)
        });
        const cpSizeTargetNow = () => {
          if (knobEntry("cp").freeSize) return null;
          const dpIntent = value.dpAttn !== null && value.dpAttn !== undefined ? value.dpAttn : h.hasFlag(flags, "--enable-dp-attention") ? h.parseIntFlagAny(flags, h.DP_HEADS) ?? 1 : false;
          if (typeof dpIntent === "number" && dpIntent > 1) return null;
          return h.parseIntFlagAny(flags, h.TP_HEADS);
        };
        const blocked = (id, v) => {
          const facts = factsNow();
          const kc = h.evaluateChip(knobEntry(id), facts);
          if (kc.hidden || kc.disabled) return true;
          if (id === "cp" && typeof v === "number" && v > 1) {
            const target = cpSizeTargetNow();
            if (target !== null && v !== target) return true;
          }
          const e = h.findEntry(knobEntry(id).values || [], v);
          return !!(e !== null && e !== undefined && h.evaluateChip(e, facts).disabled);
        };
        if (value.tp !== null && !blocked("tp", value.tp)) {
          const tpHead = h.flagSpelling(flags, h.TP_HEADS, "--tp");
          flags = h.stripFlagsByFirstToken(flags, h.TP_HEADS);
          flags = h.insertAfter(flags, h.ANCHOR_NEAR_MODEL_PATH, [`${tpHead} ${value.tp}`]);
        }
        const cpStrategyOverride = value.cpStrategy && !blocked("cpStrategy", value.cpStrategy) ? value.cpStrategy : null;
        const cpPick = value.cp !== null ? value.cp : cpStrategyOverride && cpEnabledIn(flags) ? h.parseIntFlag(flags, "--attn-cp-size") ?? 2 : null;
        const cpStrategyPick = cpStrategyOverride || bakedCpStrategy(flags) || "interleave";
        if (cpPick !== null && !blocked("cp", cpPick)) {
          flags = h.stripFlagsByFirstToken(flags, CP_OWNED_HEADS);
          if (cpPick > 1) {
            flags = h.insertAfter(flags, h.ANCHOR_NEAR_DPATTN, [`--attn-cp-size ${cpPick}`, "--enable-prefill-cp", `--cp-strategy ${cpStrategyPick}`]);
          }
        }
        const dpForced = (knobEntry("dpAttn").forceOff || []).find(r => r && h.matchConstraint(factsNow(), r.when));
        if (dpForced) {
          flags = h.stripFlagsByFirstToken(flags, [...h.DP_HEADS, "--enable-dp-attention", "--enable-dp-attention-local-control-broadcast"]);
          const stripEnv = dpForced.stripEnv || [];
          if (stripEnv.length) {
            env = env.filter(e => !stripEnv.includes(e.split("=")[0]));
          }
        } else if (value.dpAttn !== null && value.dpAttn !== undefined && !blocked("dpAttn", value.dpAttn)) {
          const dpHead = h.flagSpelling(flags, h.DP_HEADS, "--dp-size");
          const hadLocalBroadcast = h.hasFlag(flags, "--enable-dp-attention-local-control-broadcast");
          flags = h.stripFlagsByFirstToken(flags, [...h.DP_HEADS, "--enable-dp-attention", "--enable-dp-attention-local-control-broadcast"]);
          if (typeof value.dpAttn === "number" && value.dpAttn > 0) {
            flags = h.insertAfter(flags, h.ANCHOR_NEAR_TP, [`${dpHead} ${value.dpAttn}`, "--enable-dp-attention", ...hadLocalBroadcast ? ["--enable-dp-attention-local-control-broadcast"] : []]);
          }
        }
        return {
          flags,
          env
        };
      },
      render: ({axisId, value, setValue, fc, base, s, h, renderSelect, derived}) => {
        const knobs = fc.knobs || [];
        if (!knobs.length) return null;
        const setKnob = (k, v) => setValue({
          ...value,
          [k]: v
        });
        const labelFor = knob => c => {
          if (c.label !== undefined) return c.label;
          if (knob.id === "dpAttn") {
            const labelMap = knob.labels || ({
              "auto": "Auto",
              "false": "Off"
            });
            const k = c.value === null ? "auto" : String(c.value);
            return labelMap[k] || k;
          }
          return c.value === null ? "Auto" : String(c.value);
        };
        const knobDisplay = knob => {
          const v = value[knob.id];
          if (v !== null && v !== undefined) return v;
          if (derived && derived[knob.id] !== undefined) return derived[knob.id];
          return null;
        };
        const hideNullFor = knob => {
          const d = derived ? derived[knob.id] : null;
          return d !== null && d !== undefined ? [null] : [];
        };
        const entriesFor = knob => {
          const vals = knob.values || [null];
          if (knob.id !== "cp" || knob.freeSize) return vals;
          const target = base.cpSizeTarget;
          if (target === null || target === undefined) return vals;
          return vals.map(entry => {
            const v = entry === null || typeof entry !== "object" ? entry : entry.id !== undefined ? entry.id : entry.value;
            if (typeof v !== "number" || v <= 1 || v === target) return entry;
            const wrapped = entry === null || typeof entry !== "object" ? {
              value: entry
            } : {
              ...entry
            };
            return {
              ...wrapped,
              disabled: true,
              disableReason: `SGLang derives the prefill-CP size as attn_cp_size = TP / DP-Attention (= ${target} here), so only that size can be enabled.`
            };
          });
        };
        return <div key={axisId} style={s.card}>
            <div style={s.compactRow}>
              <span style={s.axisTitle}>Attention</span>
              {knobs.map(knob => {
          const kc = h.evaluateChip(knob, base);
          if (kc.hidden) return null;
          const forced = (knob.forceOff || []).find(r => r && h.matchConstraint(base, r.when));
          return <span key={knob.id} style={s.field}>
                    <span style={s.fieldLabel}>{knob.label || knob.id.toUpperCase()}</span>
                    {renderSelect(forced ? false : knobDisplay(knob), entriesFor(knob), nv => setKnob(knob.id, nv), base, labelFor(knob), {
            hideValues: hideNullFor(knob),
            disabled: kc.disabled || !!forced,
            disabledReason: forced ? forced.reason : kc.disableReason
          })}
                  </span>;
        })}
            </div>
          </div>;
      }
    },
    moe: {
      initState: () => ({
        backend: null,
        ep: null,
        mmQuant: null
      }),
      deriveFromBase: (cell, fc, h) => {
        const flags = cell && cell.flags || [];
        const a2a = h.findFlagArg(flags, "--moe-a2a-backend");
        const runner = h.findFlagArg(flags, "--moe-runner-backend");
        const w4a4 = h.hasFlag(flags, "--enable-w4a4-mxfp4-megamoe");
        return {
          backend: a2a || runner || null,
          ep: h.parseIntFlagAny(flags, h.EP_HEADS),
          mmQuant: w4a4 ? "w4a4" : "w4a8"
        };
      },
      apply: ({flags, env, value, fc, h, derived}) => {
        if (value.backend !== null) {
          flags = h.stripFlagsByFirstToken(flags, ["--moe-a2a-backend", "--moe-runner-backend"]);
          const backendEnvKeys = [];
          for (const o of fc.backend?.options || []) {
            for (const e of o.env || []) backendEnvKeys.push(e.split("=")[0]);
          }
          if (backendEnvKeys.length) env = h.stripEnvByPrefix(env, backendEnvKeys);
          const opt = (fc.backend?.options || []).find(o => o.id === value.backend);
          if (opt?.flags?.length) {
            flags = h.insertAfter(flags, h.ANCHOR_NEAR_DPATTN, opt.flags);
          }
          if (opt?.env?.length) env = [...env, ...opt.env];
        }
        const mq = fc.megamoeQuant;
        if (mq) {
          const quantKeys = [];
          const quantFlagHeads = [];
          for (const o of mq.options || []) {
            for (const e of o.env || []) quantKeys.push(e.split("=")[0]);
            for (const f of o.flags || []) quantFlagHeads.push(f.split(/[\s=]/)[0]);
          }
          flags = h.stripFlagsByFirstToken(flags, quantFlagHeads);
          const effBackend = value.backend !== null ? value.backend : derived && derived.backend;
          if (effBackend === "megamoe") {
            env = h.stripEnvByPrefix(env, [...mq.stripEnv || [], ...quantKeys]);
            const quant = value.mmQuant != null ? value.mmQuant : derived && derived.mmQuant || "w4a8";
            const opt = (mq.options || []).find(o => o.id === quant);
            if (opt?.flags?.length) {
              flags = h.insertAfter(flags, h.ANCHOR_NEAR_MOE, opt.flags);
            }
            if (opt?.env?.length) env = [...env, ...opt.env];
          } else if (value.backend !== null) {
            env = h.stripEnvByPrefix(env, quantKeys);
          }
        }
        if (value.ep !== null) {
          const epHead = h.flagSpelling(flags, h.EP_HEADS, "--ep");
          flags = h.stripFlagsByFirstToken(flags, h.EP_HEADS);
          if (value.ep > 1) {
            flags = h.insertAfter(flags, h.ANCHOR_NEAR_MOE, [`${epHead} ${value.ep}`]);
          }
        }
        return {
          flags,
          env
        };
      },
      render: ({axisId, value, setValue, fc, base, s, renderSelect, derived}) => {
        if (!fc.backend && !fc.ep) return null;
        const setSlot = (k, v) => setValue({
          ...value,
          [k]: v
        });
        const slotDisplay = k => {
          const v = value[k];
          if (v !== null && v !== undefined) return v;
          if (derived && derived[k] !== undefined) return derived[k];
          return null;
        };
        const hideNull = k => {
          const d = derived ? derived[k] : null;
          return d !== null && d !== undefined ? [null] : [];
        };
        const mmOpt = (fc.backend?.options || []).find(o => o.id === "megamoe");
        const mmAvail = !!mmOpt && (!mmOpt.requiresHw || mmOpt.requiresHw.includes(base.hw)) && (!mmOpt.excludesStrategy || !mmOpt.excludesStrategy.includes(base.strategy));
        const backendIsMega = slotDisplay("backend") === "megamoe";
        const epShown = !!fc.ep && !(typeof fc.ep.showWhen === "function" && !fc.ep.showWhen(base));
        return <div key={axisId} style={s.card}>
            <div style={s.compactRow}>
              <span style={s.axisTitle}>MoE</span>
              {fc.backend && <span style={s.field}>
                  <span style={s.fieldLabel}>Backend</span>
                  {renderSelect(slotDisplay("backend"), fc.backend.options || [], v => setSlot("backend", v), base, undefined, {
          hideValues: [...hideNull("backend"), ...mmAvail ? [] : ["megamoe"]]
        })}
                </span>}
              {fc.megamoeQuant && backendIsMega && <span style={s.field}>
                  <span style={s.fieldLabel}>Quantization</span>
                  {renderSelect(value.mmQuant != null ? value.mmQuant : derived && derived.mmQuant || "w4a8", fc.megamoeQuant.options || [], v => setSlot("mmQuant", v), base)}
                </span>}
              {epShown && <span style={s.field}>
                  <span style={s.fieldLabel}>{fc.ep.label || "EP"}</span>
                  {renderSelect(slotDisplay("ep"), fc.ep.values || [null], v => setSlot("ep", v), base, undefined, {
          hideValues: hideNull("ep")
        })}
                </span>}
            </div>
          </div>;
      }
    },
    parsers: {
      initState: fc => {
        const out = {};
        for (const item of fc.items || []) out[item.id] = null;
        return out;
      },
      deriveFromBase: (cell, fc, h) => {
        const flags = cell && cell.flags || [];
        const out = {};
        for (const item of fc.items || []) {
          const prefix = item.flag.split(/[\s=]/)[0];
          out[item.id] = h.hasFlag(flags, prefix);
        }
        return out;
      },
      apply: ({flags, env, value, fc, h, derived}) => {
        const items = fc.items || [];
        const eff = {};
        const baseOf = {};
        for (const item of items) {
          baseOf[item.id] = derived ? !!derived[item.id] : false;
          const v = value[item.id];
          eff[item.id] = v === null || v === undefined ? baseOf[item.id] : v;
        }
        const anyOverride = items.some(it => eff[it.id] !== baseOf[it.id]);
        if (!anyOverride) return {
          flags,
          env
        };
        flags = h.stripFlagsByFirstToken(flags, ["--reasoning-parser", "--tool-call-parser"]);
        const adds = [];
        for (const item of items) {
          if (eff[item.id]) adds.push(item.flag);
        }
        if (adds.length) flags = h.insertBeforeTail(flags, adds);
        return {
          flags,
          env
        };
      },
      render: ({axisId, value, setValue, fc, base, s, h, renderChip, derived}) => {
        const visible = (fc.items || []).map(item => ({
          item,
          c: h.evaluateChip(item, base)
        })).filter(({c}) => !c.hidden);
        if (visible.length === 0) return null;
        const effOn = id => {
          const v = value[id];
          if (v !== null && v !== undefined) return v;
          if (derived && derived[id] !== undefined) return derived[id];
          return false;
        };
        return <div key={axisId} style={s.card}>
            <div style={s.compactRow}>
              <span style={s.axisTitle}>Parsers</span>
              {visible.map(({item, c}) => <span key={item.id} style={s.field}>
                  {renderChip(item.label, effOn(item.id), true, () => setValue({
          ...value,
          [item.id]: !effOn(item.id)
        }), {
          disabled: c.disabled,
          disabledReason: c.disableReason
        })}
                </span>)}
            </div>
          </div>;
      }
    },
    speculative: {
      initState: () => "current",
      deriveFromBase: (cell, fc) => {
        const flags = cell && cell.flags || [];
        const baseSpec = flags.filter(f => {
          const head = f.split(/[\s=]/)[0];
          return head === "--speculative-algorithm" || head === "--speculative-num-steps" || head === "--speculative-eagle-topk" || head === "--speculative-num-draft-tokens" || head === "--speculative-adaptive" || head === "--speculative-dspark-block-size" || head === "--enable-linear-replayssm-spec" || head === "--linear-replayssm-cache-len" || head === "--speculative-ngram-max-bfs-breadth";
        });
        if (baseSpec.length === 0) return "off";
        for (const opt of fc.options || []) {
          if (!opt.flags || opt.flags.length !== baseSpec.length) continue;
          const ok = opt.flags.every(pf => baseSpec.includes(pf));
          if (ok) return opt.id;
        }
        return "current";
      },
      apply: ({flags, env, value, fc, sel, h, derived}) => {
        if (value === "current") return {
          flags,
          env
        };
        if (derived && value === derived) return {
          flags,
          env
        };
        const picked = (fc.options || []).find(p => p.id === value);
        if (picked && h.evaluateChip(picked, {
          ...sel,
          dpAttnOn: h.hasFlag(flags, "--enable-dp-attention")
        }).disabled) {
          return {
            flags,
            env
          };
        }
        flags = h.stripFlagsByFirstToken(flags, ["--speculative-algorithm", "--speculative-num-steps", "--speculative-eagle-topk", "--speculative-num-draft-tokens", "--speculative-adaptive", "--speculative-dspark-block-size", "--enable-linear-replayssm-spec", "--linear-replayssm-cache-len", "--speculative-ngram-max-bfs-breadth"]);
        const preset = (fc.options || []).find(p => p.id === value);
        if (preset?.flags?.length) flags = h.insertBeforeTail(flags, preset.flags);
        return {
          flags,
          env
        };
      },
      render: ({axisId, value, setValue, fc, base, s, h, renderChip, derived}) => {
        const opts = fc.options || [];
        if (!opts.length) return null;
        const display = value !== "current" ? value : derived ? derived : "current";
        const hideCurrent = !!(derived && derived !== "current");
        const visible = opts.map(opt => h.evaluateChip(opt, base)).filter(c => !c.hidden && !(hideCurrent && c.value === "current"));
        if (visible.length === 0) return null;
        const note = (visible.find(c => c.value === display) || ({})).note;
        return <div key={axisId} style={s.card}>
            <div style={s.compactRow}>
              <span style={s.axisTitle}>Speculative</span>
              {visible.map(c => <span key={c.value} style={s.field}>
                  {renderChip(c.label, display, c.value, () => setValue(c.value), {
          disabled: c.disabled,
          disabledReason: c.disableReason
        })}
                </span>)}
            </div>
            {note && <div style={s.axisNote}>{note}</div>}
          </div>;
      }
    },
    pdDisagg: {
      initState: fc => ({
        mode: "off",
        transferBackend: (fc && (fc.transferBackends || [])[0] || ({})).id || "mooncake",
        ibDevice: "auto"
      }),
      apply: ({flags, env, value, sel, fc, h}) => {
        const bootstrapPort = h.findFlagArg(flags, "--disaggregation-bootstrap-port");
        flags = h.stripFlagsByFirstToken(flags, ["--disaggregation-mode", "--disaggregation-transfer-backend", "--disaggregation-ib-device", "--disaggregation-bootstrap-port"]);
        const backends = fc.transferBackends || [];
        const mode = (fc.modes || []).length ? value.mode : sel && sel.pdMode || "off";
        if (mode === "prefill" || mode === "decode") {
          const specAlgorithm = (h.findFlagArg(flags, "--speculative-algorithm") || "").toUpperCase();
          if ((fc.incompatibleSpeculativeAlgorithms || []).includes(specAlgorithm)) {
            flags = flags.filter(flag => !flag.split(/[\s=]/)[0].startsWith("--speculative-"));
          }
          const backend = value.transferBackend || (backends[0] || ({})).id || "mooncake";
          const adds = [`--disaggregation-mode ${mode}`, `--disaggregation-transfer-backend ${backend}`];
          if (bootstrapPort) {
            adds.push(`--disaggregation-bootstrap-port ${bootstrapPort}`);
          }
          if (value.ibDevice && value.ibDevice !== "auto") {
            adds.push(`--disaggregation-ib-device ${value.ibDevice}`);
          }
          const modeMeta = (fc.modes || []).find(m => m.id === mode);
          const roleOverride = (fc.roleOverrides || []).find(r => r && r.mode === mode && r.when && h.matchConstraint(sel, r.when));
          const roleSpec = roleOverride || modeMeta;
          const modeGate = roleOverride ? null : modeMeta && modeMeta.when;
          const modeOk = !modeGate || Object.keys(modeGate).every(k => (modeGate[k] || []).includes(sel[k]));
          if (modeOk && roleSpec && roleSpec.flags && roleSpec.flags.length) {
            flags = h.stripFlagsByFirstToken(flags, roleSpec.flags.map(f => f.split(/[\s=]/)[0]));
            adds.push(...roleSpec.flags);
          }
          flags = h.insertBeforeTail(flags, adds);
          const servePort = PD_PORTS[mode].serve;
          flags = flags.map(f => f.split(/[\s=]/)[0] === "--port" ? `--port ${servePort}` : f);
          const meta = backends.find(b => b.id === backend);
          if (meta && meta.env && meta.env.length) {
            const gate = meta.envWhen;
            const ok = !gate || Object.keys(gate).every(k => (gate[k] || []).includes(sel[k]));
            if (ok) env = [...env, ...meta.env.filter(e => !env.includes(e))];
          }
          if (modeOk && roleSpec && roleSpec.env && roleSpec.env.length) {
            env = [...env, ...roleSpec.env.filter(e => !env.includes(e))];
          }
        }
        return {
          flags,
          env
        };
      },
      getRenderHints: (value, fc, context) => {
        const specAlgorithm = (context.h.findFlagArg(context.flags, "--speculative-algorithm") || "").toUpperCase();
        if ((fc.incompatibleSpeculativeAlgorithms || []).includes(specAlgorithm)) {
          return null;
        }
        if (value.mode === "prefill" || value.mode === "decode") {
          return {
            pdMode: value.mode
          };
        }
        return null;
      },
      render: ({axisId, value, setValue, fc, base, s, h, renderSelect}) => {
        const setSlot = (k, v) => setValue({
          ...value,
          [k]: v
        });
        const showModes = (fc.modes || []).length > 0;
        const showBackends = (fc.transferBackends || []).length > 0;
        const showIb = (fc.ibDevices || []).length > 0;
        if (!showModes && !showBackends && !showIb) return null;
        const note = (fc.notes || []).find(n => n && (!n.mode || [].concat(n.mode).includes(value.mode)) && h.matchConstraint(base, n.when));
        return <div key={axisId} style={s.card}>
            <div style={s.compactRow}>
              <span style={s.axisTitle}>PD Disagg</span>
              {showModes && <span style={s.field}>
                  <span style={s.fieldLabel}>Mode</span>
                  {renderSelect(value.mode, fc.modes, v => setSlot("mode", v), base)}
                </span>}
              {showBackends && <span style={s.field}>
                  <span style={s.fieldLabel}>Transfer Backend</span>
                  {renderSelect(value.transferBackend, fc.transferBackends, v => setSlot("transferBackend", v), base)}
                </span>}
              {showIb && <span style={s.field}>
                  <span style={s.fieldLabel}>IB Device</span>
                  {renderSelect(value.ibDevice, fc.ibDevices, v => setSlot("ibDevice", v), base)}
                </span>}
            </div>
            {note && <div style={s.axisNote}>{note.text}</div>}
          </div>;
      }
    },
    hisparse: {
      initState: fc => ({
        enable: false,
        hostRatio: fc && fc.defaultHostRatio || null
      }),
      apply: ({flags, env, value, fc, h}) => {
        const ownedHeads = ["--enable-hisparse", "--hisparse-config", ...(fc.requiredFlags || []).map(f => f.split(/\s/)[0])];
        flags = h.stripFlagsByFirstToken(flags, ownedHeads);
        const isDecode = flags.includes("--disaggregation-mode decode");
        if (value.enable && isDecode) {
          const ratio = value.hostRatio !== null && value.hostRatio !== undefined ? value.hostRatio : fc.defaultHostRatio || 10;
          const cfg = {
            ...fc.config || ({}),
            host_to_device_ratio: ratio
          };
          const adds = [...fc.requiredFlags || [], "--enable-hisparse", `--hisparse-config '${JSON.stringify(cfg)}'`];
          flags = h.insertBeforeTail(flags, adds);
        }
        return {
          flags,
          env
        };
      },
      render: ({axisId, value, setValue, fc, base, s, renderChip, renderSelect}) => {
        if (base.pdMode !== "decode") return null;
        const setSlot = (k, v) => setValue({
          ...value,
          [k]: v
        });
        const hasRatios = (fc.hostRatios || []).length > 0;
        return <div key={axisId} style={s.card}>
            <div style={s.compactRow}>
              <span style={s.axisTitle}>HiSparse</span>
              {typeof fc.showWhen !== "function" && <span style={s.field}>
                  {renderChip("Enable", value.enable, true, () => setSlot("enable", !value.enable))}
                </span>}
              {hasRatios && <span style={s.field}>
                  <span style={s.fieldLabel}>Host ratio</span>
                  {renderSelect(value.hostRatio, fc.hostRatios, v => setSlot("hostRatio", v), base)}
                </span>}
            </div>
          </div>;
      }
    },
    hicache: {
      initState: () => ({
        enable: null,
        backend: null,
        writePolicy: "auto"
      }),
      deriveFromBase: (cell, fc, h) => {
        const flags = cell && cell.flags || [];
        return {
          enable: h.hasFlag(flags, "--enable-hierarchical-cache"),
          backend: h.findFlagArg(flags, "--hicache-storage-backend"),
          writePolicy: h.findFlagArg(flags, "--hicache-write-policy") || "auto"
        };
      },
      apply: ({flags, env, value, fc, sel, h, derived}) => {
        if (fc.excludesHw && sel && fc.excludesHw.includes(sel.hw)) return {
          flags,
          env
        };
        if (typeof fc.showWhen === "function") {
          const set = (name, val) => {
            flags = h.stripFlagsByFirstToken(flags, [name]);
            if (val) flags = h.insertBeforeTail(flags, [`${name} ${val}`]);
          };
          if (value.backend) set("--hicache-storage-backend", value.backend);
          if (value.writePolicy && value.writePolicy !== "auto") {
            set("--hicache-write-policy", value.writePolicy);
          }
          return {
            flags,
            env
          };
        }
        const hasOverride = value.enable !== null || value.backend !== null || value.writePolicy && value.writePolicy !== "auto";
        if (!hasOverride) return {
          flags,
          env
        };
        const backendOptions = fc.backends || [];
        const ownedHeads = [...HICACHE_HEADS, ...(fc.requiredFlags || []).map(f => f.split(/\s/)[0]), ...backendOptions.flatMap(o => (o.flags || []).map(f => f.split(/\s/)[0]))];
        const ownedEnvKeys = [...fc.requiredEnv || [], ...backendOptions.flatMap(o => o.env || [])].map(e => e.split("=")[0]);
        flags = h.stripFlagsByFirstToken(flags, ownedHeads);
        if (ownedEnvKeys.length) env = h.stripEnvByPrefix(env, ownedEnvKeys);
        const enabled = value.enable !== null ? value.enable : !!(derived && derived.enable);
        const backend = value.backend !== null ? value.backend : derived && derived.backend || fc.defaultBackend || null;
        if (enabled) {
          const isAmd = sel && (/^mi\d/).test(sel.hw);
          const pdMode = h.findFlagArg(flags, "--disaggregation-mode") || "off";
          const pdBackend = h.findFlagArg(flags, "--disaggregation-transfer-backend");
          const roleOverride = (fc.roleOverrides || []).find(item => {
            if (!item || item.mode !== pdMode) return false;
            if (item.transferBackend && item.transferBackend !== pdBackend) return false;
            return !item.when || h.matchConstraint(sel, item.when);
          });
          const amdIo = roleOverride || isAmd && fc.amdIo;
          const ratio = amdIo && amdIo.ratio || 2;
          const useAmdIo = isAmd && amdIo;
          const adds = ["--enable-hierarchical-cache", `--hicache-ratio ${ratio}`];
          if (!useAmdIo) {
            adds.push("--hicache-size 0");
          }
          if (useAmdIo) {
            adds.push(`--hicache-mem-layout ${amdIo.memLayout}`, `--hicache-io-backend ${amdIo.ioBackend}`);
          } else if (backend) {
            adds.push("--hicache-mem-layout page_first_direct", "--hicache-io-backend direct");
          }
          const writePolicy = value.writePolicy && value.writePolicy !== "auto" ? value.writePolicy : amdIo && amdIo.writePolicy || "write_through";
          adds.push(`--hicache-write-policy ${writePolicy}`);
          if (isAmd && fc.amdStorageFileOnly ? backend === "file" : !!backend) {
            adds.push(`--hicache-storage-backend ${backend}`, `--hicache-storage-prefetch-policy ${amdIo && amdIo.prefetchPolicy || "wait_complete"}`);
          } else if (amdIo && amdIo.prefetchPolicy) {
            adds.push(`--hicache-storage-prefetch-policy ${amdIo.prefetchPolicy}`);
          }
          const backendOption = backendOptions.find(o => o.id === backend);
          adds.push(...backendOption?.flags || [], ...fc.requiredFlags || []);
          flags = h.insertBeforeTail(flags, adds);
          env = [...env, ...backendOption?.env || [], ...fc.requiredEnv || []];
        }
        return {
          flags,
          env
        };
      },
      render: ({axisId, value, setValue, fc, base, s, renderChip, renderSelect, derived}) => {
        if (fc.excludesHw && fc.excludesHw.includes(base.hw)) return null;
        const setSlot = (k, v) => setValue({
          ...value,
          [k]: v
        });
        const hasBackends = (fc.backends || []).length > 0;
        const hasPolicies = (fc.writePolicies || []).length > 0;
        const enabled = value.enable !== null ? value.enable : !!(derived && derived.enable);
        const hasAutoBackend = (fc.backends || []).some(o => o.id === null);
        const backend = value.backend !== null ? value.backend : hasAutoBackend ? null : derived && derived.backend || fc.defaultBackend || null;
        const writePolicy = value.writePolicy !== "auto" ? value.writePolicy : derived && derived.writePolicy || "auto";
        return <div key={axisId} style={s.card}>
            <div style={s.compactRow}>
              <span style={s.axisTitle}>HiCache</span>
              {typeof fc.showWhen !== "function" && <span style={s.field}>
                  {renderChip("Enable", enabled, true, () => setSlot("enable", !enabled))}
                </span>}
              {hasBackends && <span style={s.field}>
                  <span style={s.fieldLabel}>Storage</span>
                  {renderSelect(backend, fc.backends, v => setSlot("backend", v), base)}
                </span>}
              {hasPolicies && <span style={s.field}>
                  <span style={s.fieldLabel}>Write</span>
                  {renderSelect(writePolicy, fc.writePolicies, v => setSlot("writePolicy", v), base)}
                </span>}
            </div>
          </div>;
      }
    },
    umbp: {
      initState: () => ({
        enable: null,
        backend: null
      }),
      deriveFromBase: (cell, fc, h) => {
        const flags = cell && cell.flags || [];
        return {
          enable: h.hasFlag(flags, "--enable-unified-cache-external-linker"),
          backend: h.findFlagArg(flags, "--unified-cache-external-linker-backend")
        };
      },
      apply: ({flags, env, value, fc, sel, h, derived}) => {
        const ownedHeads = ["--enable-unified-cache-external-linker", "--unified-cache-external-linker-backend", ...(fc.requiredFlags || []).map(f => f.split(/\s/)[0])];
        flags = h.stripFlagsByFirstToken(flags, ownedHeads);
        if (fc.requiredEnv && fc.requiredEnv.length) {
          env = h.stripEnvByPrefix(env, fc.requiredEnv.map(e => e.split("=")[0]));
        }
        const enabled = value.enable !== null ? value.enable : !!(derived && derived.enable);
        if (!enabled) return {
          flags,
          env
        };
        if (fc.onlyHw && sel && !fc.onlyHw.includes(sel.hw)) return {
          flags,
          env
        };
        if (fc.requiresDpAttention && !flags.some(f => f.split(/[\s=]/)[0] === "--enable-dp-attention")) {
          return {
            flags,
            env
          };
        }
        flags = h.stripFlagsByFirstToken(flags, HICACHE_HEADS);
        const backend = value.backend || derived && derived.backend || fc.defaultBackend || "mori";
        flags = h.insertBeforeTail(flags, ["--enable-unified-cache-external-linker", `--unified-cache-external-linker-backend ${backend}`, ...fc.requiredFlags || []]);
        env = [...env, ...(fc.requiredEnv || []).filter(e => !env.includes(e))];
        return {
          flags,
          env
        };
      },
      render: ({axisId, value, setValue, fc, base, s, renderChip, renderSelect, derived}) => {
        if (fc.onlyHw && !fc.onlyHw.includes(base.hw)) return null;
        const setSlot = (k, v) => setValue({
          ...value,
          [k]: v
        });
        const enabled = value.enable !== null ? value.enable : !!(derived && derived.enable);
        const needsDp = !!fc.requiresDpAttention && !base.dpAttnOn;
        const backend = value.backend !== null ? value.backend : derived && derived.backend || fc.defaultBackend || "mori";
        return <div key={axisId} style={s.card}>
            <div style={s.compactRow}>
              <span style={s.axisTitle}>UMBP</span>
              <span style={s.field}>
                {renderChip("Enable", enabled, true, () => setSlot("enable", !enabled), {
          disabled: needsDp,
          disabledReason: needsDp ? "Needs DP Attention — the linker keyspace is per DP rank, so under pure TP the store holds one copy per TP rank." : ""
        })}
              </span>
              {(fc.backends || []).length > 0 && <span style={s.field}>
                  <span style={s.fieldLabel}>Store</span>
                  {renderSelect(backend, fc.backends, v => setSlot("backend", v), base)}
                </span>}
            </div>
          </div>;
      }
    },
    flagSelects: {
      initState: (fc, base) => {
        const out = {};
        for (const spec of fc || []) {
          const d = typeof spec.default === "function" ? spec.default(base) : spec.default;
          out[spec.id] = d ?? null;
        }
        return out;
      },
      deriveFromBase: (cell, fc) => {
        const flags = cell && cell.flags || [];
        const out = {};
        for (const spec of fc || []) {
          const prefixes = spec.stripPrefixes || [];
          const fam = flags.filter(f => prefixes.includes(f.split(/[\s=]/)[0]));
          let hit = null;
          for (const opt of spec.options || []) {
            if (typeof opt.flags === "function") continue;
            const of = opt.flags || [];
            if (of.length === fam.length && of.every(x => fam.includes(x))) {
              hit = opt.id;
              break;
            }
          }
          out[spec.id] = hit;
        }
        return out;
      },
      apply: ({flags, env, value, fc, sel, h, derived}) => {
        const evalBase = {
          ...sel || ({}),
          dpAttnOn: h.hasFlag(flags, "--enable-dp-attention"),
          pdMode: h.findFlagArg(flags, "--disaggregation-mode") || "off"
        };
        for (const spec of fc || []) {
          if (typeof spec.showWhen === "function" && !spec.showWhen(sel, value, derived)) continue;
          const v = value ? value[spec.id] : null;
          if (v === null || v === undefined) continue;
          const d = derived ? derived[spec.id] : null;
          if (v === d) continue;
          const opt = (spec.options || []).find(o => o.id === v);
          if (!opt) continue;
          if (h.evaluateChip(opt, evalBase).disabled) continue;
          const optFlags = typeof opt.flags === "function" ? opt.flags(value, evalBase) : opt.flags || [];
          if (optFlags === null) continue;
          const strip = new Set(spec.stripPrefixes || []);
          const byTok = new Map();
          for (const f of optFlags) {
            const t = f.split(/[\s=]/)[0];
            if (!byTok.has(t)) byTok.set(t, []);
            byTok.get(t).push(f);
          }
          const consumed = new Set();
          const next = [];
          for (const f of flags) {
            const t = f.split(/[\s=]/)[0];
            if (byTok.has(t)) {
              if (!consumed.has(t)) {
                next.push(...byTok.get(t));
                consumed.add(t);
              }
            } else if (!strip.has(t)) {
              next.push(f);
            }
          }
          const fresh = [];
          for (const [t, fs] of byTok) {
            if (!consumed.has(t)) fresh.push(...fs);
          }
          flags = fresh.length ? h.insertBeforeTail(next, fresh) : next;
          const envKeys = [...spec.stripEnv || []];
          for (const o of spec.options || []) {
            for (const e of o.env || []) envKeys.push(e.split("=")[0]);
          }
          if (envKeys.length) env = h.stripEnvByPrefix(env, envKeys);
          if (opt.env && opt.env.length) env = [...env, ...opt.env];
        }
        return {
          flags,
          env
        };
      },
      render: ({axisId, value, setValue, fc, base, s, h, renderChip, derived}) => {
        const cards = [];
        for (const spec of fc || []) {
          if (typeof spec.showWhen === "function" && !spec.showWhen(base, value, derived)) continue;
          const opts = (spec.options || []).map(o => h.evaluateChip(o, base)).filter(c => !c.hidden);
          if (!opts.length) continue;
          const explicit = value ? value[spec.id] : null;
          const display = explicit !== null && explicit !== undefined ? explicit : derived ? derived[spec.id] : null;
          if (spec.control === "slider") {
            const idx = Math.max(0, opts.findIndex(c => c.value === display));
            const cur = opts[idx];
            cards.push(<div key={`${axisId}-${spec.id}`} style={s.card}>
                <div style={s.compactRow}>
                  <span style={s.axisTitle}>{spec.title}</span>
                  <input type="range" min={0} max={opts.length - 1} step={1} value={idx} onChange={e => setValue({
              ...value,
              [spec.id]: opts[Number(e.target.value)].value
            })} style={{
              flex: 1,
              minWidth: "120px",
              accentColor: "#D45D44"
            }} />
                  <span style={{
              ...s.axisTitle,
              minWidth: "24px",
              textAlign: "right"
            }}>
                    {cur ? cur.label : "-"}
                  </span>
                </div>
              </div>);
            continue;
          }
          cards.push(<div key={`${axisId}-${spec.id}`} style={s.card}>
              <div style={s.compactRow}>
                <span style={s.axisTitle}>{spec.title}</span>
                {opts.map(c => <span key={c.value} style={s.field}>
                    {renderChip(c.label, display, c.value, () => setValue({
            ...value,
            [spec.id]: c.value
          }), {
            disabled: c.disabled,
            disabledReason: c.disableReason
          })}
                  </span>)}
              </div>
            </div>);
        }
        return cards.length ? cards : null;
      }
    }
  };
  const applyAllDeltas = (baseFlags, baseEnv, allDeltas, sel, derivedMap) => {
    let flags = [...baseFlags];
    let env = [...baseEnv || []];
    let pdMode = null;
    const pdFc = pgFeatures.pdDisagg;
    const pdDelta = allDeltas.pdDisagg;
    const pdRoleSel = pdFc && (pdFc.modes || []).length && pdDelta ? pdDelta.mode : sel && sel.pdMode || "off";
    for (const [axisId, handler] of Object.entries(AXIS_HANDLERS)) {
      const fc = pgFeatures[axisId];
      if (!fc) continue;
      const value = allDeltas[axisId];
      if (value === undefined) continue;
      const derived = derivedMap ? derivedMap[axisId] : null;
      const specAlgorithm = (findFlagArg(flags, "--speculative-algorithm") || "").toUpperCase() || null;
      const liveSel = {
        ...sel,
        specAlgorithm,
        pdMode: pdRoleSel
      };
      const out = handler.apply({
        flags,
        env,
        value,
        fc,
        sel: liveSel,
        h: helpers,
        derived
      });
      flags = out.flags;
      env = out.env;
      if (handler.getRenderHints) {
        const hints = handler.getRenderHints(value, fc, {
          flags,
          env,
          sel: liveSel,
          h: helpers
        }) || ({});
        if (hints.pdMode) pdMode = hints.pdMode;
      }
    }
    return {
      flags,
      env,
      pdMode
    };
  };
  const renderCommandLines = (cell, flags, cellEnv, sel, envValues, pdMode = null, mode = "python") => {
    const modelName = resolveModelName(sel);
    let f = [...flags];
    const nnodesFlag = f.find(x => x.split(/[\s=]/)[0] === "--nnodes");
    const nnodesMatch = nnodesFlag && (/^--nnodes(?:\s+|=)(\d+)$/).exec(nnodesFlag.trim());
    const baseNnodes = sel.nodes !== undefined ? parseNnodes(sel.nodes) : cell && cell.nnodes || 1;
    const nnodes = nnodesMatch ? parseInt(nnodesMatch[1], 10) : baseNnodes;
    const multinode = nnodes > 1;
    if (multinode && !f.some(x => x.startsWith("--nnodes"))) {
      const PARALLELISM_ANCHORS = ["--enable-dp-attention", "--dp-size", "--dp", "--tp-size", "--tp"];
      let at = -1;
      for (const anchor of PARALLELISM_ANCHORS) {
        at = f.findIndex(x => x.split(/[\s=]/)[0] === anchor);
        if (at !== -1) break;
      }
      if (at === -1) at = f.findIndex(x => x.startsWith("--model-path"));
      const distPort = pdMode && PD_PORTS[pdMode] ? PD_PORTS[pdMode].dist : 20000;
      f.splice(at + 1, 0, `--nnodes ${nnodes}`, `--node-rank {{NODE_RANK}}`, `--dist-init-addr {{NODE0_IP}}:${distPort}`);
    }
    let cmd;
    if (mode === "docker") {
      const di = config.dockerImages || ({});
      const image = di[`${sel.hw}|${sel.variant}|${sel.quant}`] || di[`${sel.variant}|${sel.quant}`] || di[`${sel.hw}|${sel.quant}|${sel.strategy}`] || di[`${sel.hw}|${sel.quant}`] || di[sel.hw] || "lmsysorg/sglang:dev";
      const dockerRunCommand = typeof config.dockerRunCommand === "function" ? config.dockerRunCommand(sel) : config.dockerRunCommand || "sglang serve";
      const portFlag = f.find(x => x.split(/[\s=]/)[0] === "--port");
      const servePort = portFlag ? portFlag.slice(("--port").length).trim() : "{{PORT}}";
      const hostNetwork = multinode || pdMode || typeof config.dockerHostNetworkWhen === "function" && config.dockerHostNetworkWhen(sel, {
        flags: f,
        env: cellEnv
      });
      const AMD_RDMA_DOCKER_FLAGS = ["--device /dev/infiniband", "--cap-add IPC_LOCK", "--ulimit memlock=-1", "--ulimit stack=67108864", "--ulimit nofile=1048576:1048576"];
      const HW_MULTINODE_DOCKER_FLAGS = {
        "dgx-spark": ["--ulimit memlock=-1:-1", "--cap-add IPC_LOCK", "--device /dev/infiniband"],
        mi300x: AMD_RDMA_DOCKER_FLAGS,
        mi325x: AMD_RDMA_DOCKER_FLAGS,
        mi350x: AMD_RDMA_DOCKER_FLAGS,
        mi355x: AMD_RDMA_DOCKER_FLAGS
      };
      const fabricFlags = HW_MULTINODE_DOCKER_FLAGS[sel.hw] || [];
      const isAmdHw = (/^mi\d/).test(sel.hw || "");
      const dockerLines = [...isAmdHw ? ["docker run", "  --device=/dev/kfd --device=/dev/dri", "  --group-add video", "  --cap-add=SYS_PTRACE --security-opt seccomp=unconfined", "  --shm-size 32g"] : ["docker run --gpus all", "  --shm-size 32g"], hostNetwork ? "  --network host" : `  -p ${servePort}:${servePort}`, ...multinode || pdMode ? fabricFlags.map(x => "  " + x) : [], "  -v ~/.cache/huggingface:/root/.cache/huggingface", ...(config.dockerMounts || []).map(mount => `  -v ${mount}`), `  --env "HF_TOKEN={{HF_TOKEN}}"`, ...cellEnv.map(e => `  --env ${e}`), "  --ipc=host", `  ${image}`, `  ${dockerRunCommand}`, ...f.map(x => "    " + x)];
      cmd = dockerLines.join(" \\\n");
    } else {
      const flagBlock = f.map(x => "  " + x).join(" \\\n");
      const envBlock = cellEnv.length ? cellEnv.join(" \\\n") + " \\\n" : "";
      cmd = `${envBlock}sglang serve \\\n${flagBlock}`;
    }
    if (multinode && config.multiNodeHints && config.multiNodeHints[sel.hw]) {
      const hint = config.multiNodeHints[sel.hw].map(line => line.length ? "# " + line : "#").join("\n");
      cmd = `${hint}\n${cmd}`;
    }
    cmd = interpolate(cmd, envValues, modelName);
    if (multinode) {
      const header = `# Multi-node (${nnodes} nodes). Run the same command on every node with:\n` + `#   <node-rank> = 0 on the head node, 1..${nnodes - 1} on the others\n` + `#   <node0-ip>  = IP of the head node (reachable from all others)`;
      cmd = `${header}\n${cmd}`;
    }
    if (pdMode === "prefill" || pdMode === "decode") {
      const sibling = pdMode === "prefill" ? "decode" : "prefill";
      const routerCfg = resolveRouter(config.playgroundFeatures && config.playgroundFeatures.pdDisagg, sel);
      const routerPort = routerCfg && routerCfg.port || 8000;
      const routerLine = routerCfg ? `# then front BOTH with the Router shown below.\n` + `# Client traffic (cURL) targets the router (:${routerPort}), not this role server.` : `# then front BOTH with a router; client traffic targets the router, not this role server.`;
      const hicacheCfg = config.playgroundFeatures && config.playgroundFeatures.hicache;
      const pdBackend = findFlagArg(f, "--disaggregation-transfer-backend");
      const hicacheEnabled = f.some(x => x === "--enable-hierarchical-cache");
      const hicacheNotice = hicacheEnabled && hicacheCfg ? (hicacheCfg.notices || []).find(item => {
        if (!item || item.mode !== pdMode) return false;
        if (item.transferBackend && item.transferBackend !== pdBackend) return false;
        return !item.when || matchConstraint(sel, item.when);
      }) : null;
      const noticeLine = hicacheNotice && hicacheNotice.text ? `# Note: ${hicacheNotice.text}\n` : "";
      const banner = `# === PD Disaggregation: ${pdMode.toUpperCase()} role ===\n` + noticeLine + `# Runs the ${pdMode} server. Also run the ${sibling} role on its peer host,\n` + routerLine;
      cmd = `${banner}\n${cmd}`;
    }
    return cmd;
  };
  const computeDiff = (baseStr, pgStr) => {
    const a = baseStr.split("\n");
    const b = pgStr.split("\n");
    const m = a.length, n = b.length;
    const dp = Array(m + 1).fill(null).map(() => new Array(n + 1).fill(0));
    for (let i = 1; i <= m; i++) {
      for (let j = 1; j <= n; j++) {
        if (a[i - 1] === b[j - 1]) dp[i][j] = dp[i - 1][j - 1] + 1; else dp[i][j] = Math.max(dp[i - 1][j], dp[i][j - 1]);
      }
    }
    const out = [];
    let i = m, j = n;
    while (i > 0 || j > 0) {
      if (i > 0 && j > 0 && a[i - 1] === b[j - 1]) {
        out.unshift({
          line: a[i - 1],
          kind: "unchanged"
        });
        i--;
        j--;
      } else if (j > 0 && (i === 0 || dp[i][j - 1] >= dp[i - 1][j])) {
        out.unshift({
          line: b[j - 1],
          kind: "added"
        });
        j--;
      } else {
        out.unshift({
          line: a[i - 1],
          kind: "removed"
        });
        i--;
      }
    }
    return out;
  };
  const serializeCell = (sel, env, flags) => {
    const matchEntries = [`hw: ${JSON.stringify(sel.hw)}`, `variant: ${JSON.stringify(sel.variant)}`, `quant: ${JSON.stringify(sel.quant)}`, `strategy: ${JSON.stringify(sel.strategy)}`, `nodes: ${JSON.stringify(sel.nodes)}`].join(", ");
    const fmtList = items => {
      if (!items || items.length === 0) return "[]";
      const lines = items.map(s => `        ${JSON.stringify(s)},`).join("\n");
      return `[\n${lines}\n      ]`;
    };
    return ["    {", `      match: { ${matchEntries} },`, "      verified: true,", `      env: ${fmtList(env)},`, `      flags: ${fmtList(flags)},`, "    },"].join("\n");
  };
  const buildSubmitUrl = (sel, fields) => {
    const gh = config.github || ({});
    const owner = gh.owner || "sgl-project";
    const repo = gh.repo || "sglang";
    const tmpl = gh.issueTemplate || "3-playground-verified-cell.yml";
    const cookbookModel = gh.cookbookModel || "deepseek-ai/deepseek-v4";
    const combo = `${sel.hw} / ${sel.variant} / ${sel.quant} / ${sel.strategy} / ${sel.nodes}`;
    const params = new URLSearchParams({
      template: tmpl,
      title: `[Playground] Verified cell: ${combo}`,
      model: cookbookModel,
      combination: combo,
      "cell-snippet": fields.cellSnippet || "",
      "existing-cell": fields.existingCell || "",
      "sglang-version": fields.sglangVersion || "",
      "bench-result": fields.benchResult || "",
      notes: fields.notes || ""
    });
    return `https://github.com/${owner}/${repo}/issues/new?${params.toString()}`;
  };
  const makeStyles = isDark => ({
    container: {
      maxWidth: "900px",
      margin: "0 auto",
      display: "flex",
      flexDirection: "column",
      gap: "6px"
    },
    card: {
      padding: "6px 10px",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      borderLeft: `3px solid ${isDark ? "#FDBA74" : "#FB923C"}`,
      borderRadius: "4px",
      background: isDark ? "#1f2937" : "#fff"
    },
    cardStack: {
      display: "flex",
      flexDirection: "column",
      gap: "6px"
    },
    baseStrip: {
      padding: "8px 12px",
      borderRadius: "4px",
      background: isDark ? "#064e3b" : "#d1fae5",
      color: isDark ? "#a7f3d0" : "#065f46",
      fontSize: "12px",
      display: "flex",
      alignItems: "center",
      gap: "10px"
    },
    title: {
      fontSize: "13px",
      fontWeight: "600",
      color: isDark ? "#e5e7eb" : "inherit",
      marginBottom: "8px"
    },
    compactRow: {
      display: "flex",
      flexWrap: "wrap",
      alignItems: "center",
      gap: "10px",
      rowGap: "4px"
    },
    axisTitle: {
      fontSize: "12px",
      fontWeight: 700,
      color: isDark ? "#FDBA74" : "#C2410C",
      letterSpacing: "0.02em",
      minWidth: "100px",
      flexShrink: 0
    },
    field: {
      display: "inline-flex",
      alignItems: "center",
      gap: "4px"
    },
    fieldLabel: {
      fontSize: "11px",
      fontWeight: 500,
      color: isDark ? "#9ca3af" : "#6b7280"
    },
    select: {
      padding: "2px 6px",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "3px",
      fontSize: "12px",
      background: isDark ? "#111827" : "#fff",
      color: isDark ? "#e5e7eb" : "#111827",
      cursor: "pointer",
      lineHeight: "1.4"
    },
    rowFlex: {
      display: "flex",
      flexWrap: "wrap",
      gap: "6px",
      alignItems: "center",
      flex: 1
    },
    subRow: {
      display: "flex",
      alignItems: "center",
      gap: "10px"
    },
    subLabel: {
      fontSize: "11px",
      fontWeight: 600,
      color: isDark ? "#9ca3af" : "#6b7280",
      minWidth: "96px",
      flexShrink: 0,
      letterSpacing: "0.02em"
    },
    chipRow: {
      display: "flex",
      flexWrap: "wrap",
      gap: "6px",
      flex: 1
    },
    chip: {
      padding: "3px 9px",
      border: `1px solid ${isDark ? "#9ca3af" : "#d1d5db"}`,
      borderRadius: "3px",
      cursor: "pointer",
      fontSize: "12px",
      userSelect: "none",
      background: isDark ? "#374151" : "#fff",
      color: isDark ? "#e5e7eb" : "inherit",
      textAlign: "center"
    },
    chipChecked: {
      background: "#D45D44",
      color: "white",
      borderColor: "#D45D44"
    },
    chipDisabled: {
      cursor: "not-allowed",
      opacity: 0.4
    },
    commandWrap: {
      position: "relative",
      background: isDark ? "#111827" : "#f5f5f5",
      borderRadius: "6px",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      overflow: "hidden"
    },
    commandHeader: {
      display: "flex",
      flexWrap: "wrap",
      justifyContent: "space-between",
      alignItems: "center",
      gap: "6px 10px",
      padding: "6px 10px",
      borderBottom: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      background: isDark ? "#1f2937" : "#fafafa"
    },
    commandPre: {
      padding: "12px 16px",
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace",
      fontSize: "12px",
      lineHeight: "1.5",
      color: isDark ? "#e5e7eb" : "#374151",
      whiteSpace: "pre-wrap",
      overflowX: "auto",
      margin: 0
    },
    axisNote: {
      margin: "6px 0 0",
      padding: "6px 10px",
      borderRadius: "6px",
      fontSize: "11px",
      lineHeight: "1.45",
      background: isDark ? "#78350f" : "#fef3c7",
      color: isDark ? "#fde68a" : "#92400e",
      border: `1px solid ${isDark ? "#92400e" : "#fcd34d"}`
    },
    mtpWarn: {
      margin: "8px 0 0",
      padding: "8px 12px",
      borderRadius: "8px",
      fontSize: "12px",
      lineHeight: "1.45",
      background: isDark ? "#78350f" : "#fef3c7",
      color: isDark ? "#fde68a" : "#92400e",
      border: `1px solid ${isDark ? "#92400e" : "#fcd34d"}`
    },
    diffLineUnchanged: {
      display: "block"
    },
    diffLineAdded: {
      display: "block",
      background: isDark ? "rgba(16,185,129,0.15)" : "rgba(16,185,129,0.18)",
      color: isDark ? "#a7f3d0" : "#065f46",
      borderLeft: `3px solid #10b981`,
      paddingLeft: "8px",
      marginLeft: "-8px"
    },
    diffLineRemoved: {
      display: "block",
      background: isDark ? "rgba(239,68,68,0.10)" : "rgba(239,68,68,0.10)",
      color: isDark ? "#fca5a5" : "#991b1b",
      textDecoration: "line-through",
      opacity: 0.7,
      borderLeft: `3px solid #ef4444`,
      paddingLeft: "8px",
      marginLeft: "-8px"
    },
    badge: verified => ({
      display: "inline-flex",
      alignItems: "center",
      gap: "6px",
      padding: "2px 8px",
      borderRadius: "10px",
      background: verified ? isDark ? "#064e3b" : "#d1fae5" : isDark ? "#78350f" : "#fef3c7",
      color: verified ? isDark ? "#a7f3d0" : "#065f46" : isDark ? "#fde68a" : "#92400e",
      fontSize: "11px",
      fontWeight: 600
    }),
    badgeDot: verified => ({
      width: "8px",
      height: "8px",
      borderRadius: "50%",
      background: verified ? "#10b981" : "#f59e0b"
    }),
    iconButton: {
      padding: "4px 10px",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "4px",
      background: isDark ? "#1f2937" : "#fff",
      color: isDark ? "#e5e7eb" : "#374151",
      fontSize: "11px",
      fontWeight: 500,
      cursor: "pointer",
      display: "inline-flex",
      alignItems: "center",
      gap: "4px"
    },
    iconRow: {
      display: "inline-flex",
      flexWrap: "wrap",
      gap: "6px"
    },
    runModeWrap: {
      display: "inline-flex",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "10px",
      overflow: "hidden",
      fontSize: "11px",
      fontWeight: 600,
      userSelect: "none"
    },
    runModeChip: active => ({
      padding: "2px 10px",
      cursor: "pointer",
      background: active ? isDark ? "#1f2937" : "#fff" : "transparent",
      color: active ? isDark ? "#e5e7eb" : "#111827" : isDark ? "#9ca3af" : "#6b7280",
      borderRight: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`
    }),
    runModeChipLast: active => ({
      padding: "2px 10px",
      cursor: "pointer",
      background: active ? isDark ? "#1f2937" : "#fff" : "transparent",
      color: active ? isDark ? "#e5e7eb" : "#111827" : isDark ? "#9ca3af" : "#6b7280"
    }),
    headerLeft: {
      display: "inline-flex",
      flexWrap: "wrap",
      alignItems: "center",
      gap: "8px"
    },
    dialog: {
      background: isDark ? "#1f2937" : "#fff",
      color: isDark ? "#e5e7eb" : "#111827",
      borderRadius: "8px",
      padding: "20px",
      maxWidth: "720px",
      width: "92%",
      maxHeight: "calc(100vh - 80px)",
      overflowY: "auto",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      boxShadow: "0 10px 25px rgba(0,0,0,0.25)",
      margin: "auto"
    },
    modalHeader: {
      display: "flex",
      justifyContent: "space-between",
      alignItems: "center",
      marginBottom: "12px"
    },
    modalTitle: {
      fontSize: "15px",
      fontWeight: 600
    },
    modalCloseBtn: {
      background: "transparent",
      border: "none",
      color: "inherit",
      fontSize: "20px",
      cursor: "pointer",
      padding: "0 6px",
      lineHeight: 1
    },
    formField: {
      display: "flex",
      flexDirection: "column",
      gap: "4px",
      marginBottom: "10px"
    },
    formLabel: {
      fontSize: "12px",
      fontWeight: 500,
      color: isDark ? "#9ca3af" : "#4b5563"
    },
    formInput: {
      padding: "6px 10px",
      fontSize: "13px",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "4px",
      background: isDark ? "#111827" : "#fff",
      color: isDark ? "#e5e7eb" : "#111827",
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace"
    },
    sectionHeading: {
      fontSize: "12px",
      fontWeight: 600,
      textTransform: "uppercase",
      letterSpacing: "0.04em",
      color: isDark ? "#9ca3af" : "#6b7280",
      margin: "12px 0 6px 0"
    },
    primaryBtn: {
      padding: "6px 14px",
      background: isDark ? "#FDBA74" : "#FB923C",
      color: isDark ? "#7C2D12" : "white",
      border: "none",
      borderRadius: "4px",
      cursor: "pointer",
      fontSize: "13px",
      fontWeight: 500
    },
    resetBtn: {
      marginLeft: "auto",
      padding: "2px 8px",
      fontSize: "11px",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "3px",
      background: "transparent",
      color: isDark ? "#9ca3af" : "#6b7280",
      cursor: "pointer"
    },
    switchBaseBtn: {
      padding: "2px 8px",
      fontSize: "11px",
      fontWeight: 600,
      border: `1px solid ${isDark ? "#FDBA74" : "#FB923C"}`,
      borderRadius: "3px",
      background: "transparent",
      color: isDark ? "#FDBA74" : "#C2410C",
      cursor: "pointer"
    },
    matchedHint: {
      fontSize: "11px",
      color: isDark ? "#9ca3af" : "#6b7280",
      marginLeft: "8px",
      display: "inline-flex",
      alignItems: "center",
      gap: "4px"
    },
    matchedSwitchBtn: {
      marginLeft: "4px",
      background: "transparent",
      border: "none",
      padding: 0,
      color: isDark ? "#FDBA74" : "#C2410C",
      cursor: "pointer",
      fontSize: "11px",
      fontWeight: 600,
      textDecoration: "underline",
      textUnderlineOffset: "2px"
    }
  });
  const [isDark, setIsDark] = useState(false);
  useEffect(() => {
    const check = () => {
      const html = document.documentElement;
      setIsDark(html.classList.contains("dark") || html.getAttribute("data-theme") === "dark" || html.style.colorScheme === "dark");
    };
    check();
    const observer = new MutationObserver(check);
    observer.observe(document.documentElement, {
      attributes: true,
      attributeFilter: ["class", "data-theme", "style"]
    });
    return () => observer.disconnect();
  }, []);
  const [env, setEnv] = useState(() => placeholderDefaults(config.placeholders));
  useEffect(() => {
    try {
      const raw = window.localStorage.getItem(STORAGE_KEY);
      if (raw) {
        const parsed = JSON.parse(raw);
        setEnv({
          ...placeholderDefaults(config.placeholders),
          ...parsed
        });
      }
    } catch {}
  }, []);
  const saveEnv = next => {
    setEnv(next);
    try {
      window.localStorage.setItem(STORAGE_KEY, JSON.stringify(next));
    } catch {}
  };
  const overlayDefaults = () => {
    const out = {};
    for (const d of config.overlayDims || []) {
      const opts = d.options || [];
      out[d.id] = d.default !== undefined ? d.default : opts[0] && opts[0].id || "";
    }
    return out;
  };
  const baseFallback = () => ({
    ...config.cells[0].match,
    ...overlayDefaults()
  });
  const initialBaseFromHash = () => {
    const fallback = baseFallback();
    if (typeof window === "undefined") return {
      ...fallback
    };
    const raw = window.location.hash.replace(/^#/, "");
    if (!raw) return {
      ...fallback
    };
    const params = new URLSearchParams(raw);
    const out = {
      ...fallback
    };
    params.forEach((value, key) => {
      if ((key in out)) out[key] = value;
    });
    return out;
  };
  const [base, setBase] = useState(() => initialBaseFromHash());
  useEffect(() => {
    const onHash = () => setBase(initialBaseFromHash());
    const onSelEvent = e => {
      const fallback = baseFallback();
      const incoming = e && e.detail || ({});
      const next = {
        ...fallback
      };
      for (const k of Object.keys(next)) {
        if (incoming[k] !== undefined) next[k] = incoming[k];
      }
      setBase(next);
    };
    window.addEventListener("hashchange", onHash);
    window.addEventListener("sglang-deploy-sel", onSelEvent);
    return () => {
      window.removeEventListener("hashchange", onHash);
      window.removeEventListener("sglang-deploy-sel", onSelEvent);
    };
  }, []);
  const [pgRatios, setPgRatios] = useState({
    eff: null,
    base: null
  });
  useEffect(() => {
    const onRatio = e => setPgRatios({
      eff: e.detail && e.detail.ratio || null,
      base: e.detail && (e.detail.baseRatio || e.detail.ratio) || null
    });
    window.addEventListener("sglang-k3-mamba-ratio", onRatio);
    return () => window.removeEventListener("sglang-k3-mamba-ratio", onRatio);
  }, []);
  const initialDeltas = () => {
    const out = {};
    for (const [axisId, handler] of Object.entries(AXIS_HANDLERS)) {
      const fc = pgFeatures[axisId];
      if (fc) out[axisId] = handler.initState(fc, base);
    }
    return out;
  };
  const [deltas, setDeltas] = useState(initialDeltas);
  useEffect(() => {
    setDeltas(initialDeltas());
  }, [Object.keys(base).sort().map(k => `${k}=${base[k]}`).join("&")]);
  const [modal, setModal] = useState(null);
  const openDialog = el => {
    if (el && !el.open) {
      try {
        el.showModal();
      } catch {}
    }
  };
  const onDialogClick = e => {
    if (e.target !== e.currentTarget) return;
    const r = e.currentTarget.getBoundingClientRect();
    const {clientX: x, clientY: y} = e;
    if (x < r.left || x > r.right || y < r.top || y > r.bottom) setModal(null);
  };
  useEffect(() => {
    const ID = "__playground_dialog_backdrop";
    if (document.getElementById(ID)) return undefined;
    const style = document.createElement("style");
    style.id = ID;
    style.textContent = `dialog::backdrop { background: rgba(0, 0, 0, 0.5); }`;
    document.head.appendChild(style);
    return () => {
      const el = document.getElementById(ID);
      if (el) el.remove();
    };
  }, []);
  const [copied, setCopied] = useState(false);
  const [curlCopied, setCurlCopied] = useState(false);
  const [routerCopied, setRouterCopied] = useState(false);
  const [envDraft, setEnvDraft] = useState(env);
  useEffect(() => {
    if (modal === "env") setEnvDraft(env);
  }, [modal, env]);
  const [runMode, setRunMode] = useState("python");
  const [submitDraft, setSubmitDraft] = useState({
    sglangVersion: "",
    benchResult: "",
    notes: ""
  });
  const [submitAttest, setSubmitAttest] = useState({
    ranCommand: false,
    reachedReady: false,
    outputCorrect: false
  });
  useEffect(() => {
    if (modal === "submit") {
      setSubmitDraft({
        sglangVersion: "",
        benchResult: "",
        notes: ""
      });
      setSubmitAttest({
        ranCommand: false,
        reachedReady: false,
        outputCorrect: false
      });
    }
  }, [modal]);
  const s = makeStyles(isDark);
  const baseCell = withOverlay(findCell(config.cells, base), base);
  const modelName = resolveModelName(base);
  const derivedMap = {};
  if (baseCell) {
    for (const [axisId, handler] of Object.entries(AXIS_HANDLERS)) {
      const fc = pgFeatures[axisId];
      if (!fc || !handler.deriveFromBase) continue;
      derivedMap[axisId] = handler.deriveFromBase(baseCell, fc, helpers);
    }
  }
  const attnDelta = deltas.attention || ({});
  const attnDerived = derivedMap.attention || ({});
  const attnKnobs = (pgFeatures.attention || ({})).knobs || [];
  const effTp = attnDelta.tp !== null && attnDelta.tp !== undefined ? attnDelta.tp : attnDerived.tp !== undefined ? attnDerived.tp : null;
  const staleExplicit = (knobId, picked) => {
    const knob = attnKnobs.find(k => k.id === knobId);
    const e = knob ? findEntry(knob.values || [], picked) : null;
    return !!(e !== null && e !== undefined && evaluateChip(e, {
      ...base,
      effTp
    }).disabled);
  };
  const effDpAttn = attnDelta.dpAttn !== null && attnDelta.dpAttn !== undefined ? staleExplicit("dpAttn", attnDelta.dpAttn) ? null : attnDelta.dpAttn : attnDerived.dpAttn !== undefined ? attnDerived.dpAttn : null;
  const dpAttnOn = effDpAttn === true || typeof effDpAttn === "number" && effDpAttn > 0;
  const dpDegEff = typeof effDpAttn === "number" && effDpAttn > 0 ? effDpAttn : 1;
  const cpKnobFreeSize = !!(attnKnobs.find(k => k.id === "cp") || ({})).freeSize;
  const cpSizeTarget = !cpKnobFreeSize && dpDegEff === 1 && typeof effTp === "number" && effTp > 0 ? effTp : null;
  const cpSizeStale = v => typeof v === "number" && v > 1 && cpSizeTarget !== null && v !== cpSizeTarget;
  const effCp = attnDelta.cp !== null && attnDelta.cp !== undefined ? staleExplicit("cp", attnDelta.cp) || cpSizeStale(attnDelta.cp) ? null : attnDelta.cp : attnDerived.cp !== undefined ? attnDerived.cp : null;
  const cpOn = typeof effCp === "number" && effCp > 1;
  const cpStrategy = (attnDelta.cpStrategy && !staleExplicit("cpStrategy", attnDelta.cpStrategy) ? attnDelta.cpStrategy : attnDerived.cpStrategy !== undefined ? attnDerived.cpStrategy : null) || "interleave";
  const constraintEffective = baseCell ? applyAllDeltas(baseCell.flags, baseCell.env, deltas, base, derivedMap) : null;
  const pdCardOwnsMode = (pgFeatures.pdDisagg && pgFeatures.pdDisagg.modes || []).length > 0;
  const pdMode = pdCardOwnsMode ? constraintEffective && constraintEffective.pdMode || "off" : base.pdMode || "off";
  const specAlgorithm = constraintEffective ? (findFlagArg(constraintEffective.flags, "--speculative-algorithm") || "").toUpperCase() || null : null;
  const constraintBase = {
    ...base,
    dpAttnOn,
    cpOn,
    cpStrategy,
    cpSizeTarget,
    effTp,
    pdMode,
    specAlgorithm
  };
  let baseCommand = "";
  let playgroundCommand = "";
  let diffLines = [];
  let pgFlagsLatest = [];
  let pgEnvLatest = [];
  const withRatio = (fl, value) => {
    if (!value) return fl;
    if (fl.some(f => f.startsWith("--mamba-full-memory-ratio") || f.startsWith("--max-mamba-cache-size"))) return fl;
    const out = [...fl];
    const line = `--mamba-full-memory-ratio ${value}`;
    const i = out.findIndex(f => f.startsWith("--host"));
    if (i >= 0) out.splice(i, 0, line); else out.push(line);
    return out;
  };
  if (baseCell) {
    baseCommand = renderCommandLines(baseCell, withRatio(baseCell.flags, pgRatios.base), baseCell.env, base, env, null, runMode);
    const {flags: pgFlags, env: pgEnv, pdMode} = applyAllDeltas(baseCell.flags, baseCell.env, deltas, base, derivedMap);
    pgFlagsLatest = pgFlags;
    pgEnvLatest = pgEnv;
    playgroundCommand = renderCommandLines(baseCell, withRatio(pgFlags, pgRatios.eff), pgEnv, base, env, pdMode, runMode);
    diffLines = computeDiff(baseCommand, playgroundCommand);
  }
  const effectiveKey = pgFlagsLatest.join("\n") + " " + pgEnvLatest.join("\n") + " " + (baseCell ? baseCell.flags.join("\n") + " " + baseCell.env.join("\n") : "");
  useEffect(() => {
    if (typeof window === "undefined" || !baseCell) return;
    window.dispatchEvent(new CustomEvent("sglang-k3-effective-config", {
      detail: {
        flags: pgFlagsLatest,
        env: pgEnvLatest,
        baseFlags: baseCell.flags,
        baseEnv: baseCell.env
      }
    }));
  }, [effectiveKey]);
  const matchedCell = baseCell ? findMatchingCell(config.cells, base, pgEnvLatest, pgFlagsLatest) : null;
  const playgroundVerified = !!(matchedCell && matchedCell.verified);
  const matchedSiblingCell = matchedCell && DIMENSIONS.some(d => matchedCell.match[d] !== base[d]) ? matchedCell : null;
  const pgSpecAlgoFlag = pgFlagsLatest.find(f => f.split(/[\s=]/)[0] === "--speculative-algorithm");
  const pgPdRole = pdMode === "prefill" || pdMode === "decode";
  const pgSpecHint = !!pgSpecAlgoFlag && (pgPdRole || !pgFlagsLatest.some(f => f.split(/[\s=]/)[0] === "--max-running-requests"));
  const specAlgoLabels = {
    EAGLE: "MTP",
    EAGLE3: "MTP",
    FROZEN_KV_MTP: "MTP",
    DSPARK: "DSpark",
    DFLASH: "DFlash",
    NGRAM: "N-gram",
    STANDALONE: "standalone draft"
  };
  const pgSpecAlgoValue = pgSpecAlgoFlag ? pgSpecAlgoFlag.split(/[\s=]/).filter(Boolean)[1] || "" : "";
  const pgSpecAlgoName = specAlgoLabels[pgSpecAlgoValue.toUpperCase()] || pgSpecAlgoValue || "MTP";
  const pgCpDpHint = cpEnabledIn(pgFlagsLatest) && (bakedCpStrategy(pgFlagsLatest) || "interleave") === "interleave" && pgFlagsLatest.some(f => f.split(/[\s=]/)[0] === "--enable-dp-attention");
  const proposedCellSnippet = baseCell ? serializeCell(base, pgEnvLatest, pgFlagsLatest) : "";
  const existingCellSnippet = baseCell ? serializeCell(base, baseCell.env || [], baseCell.flags) : "";
  const submitUrl = baseCell ? buildSubmitUrl(base, {
    cellSnippet: proposedCellSnippet,
    existingCell: existingCellSnippet,
    sglangVersion: submitDraft.sglangVersion,
    benchResult: submitDraft.benchResult,
    notes: submitDraft.notes
  }) : "";
  const submitReady = submitAttest.ranCommand && submitAttest.reachedReady && submitAttest.outputCorrect && submitDraft.sglangVersion.trim().length > 0;
  const pdRouter = pdMode !== "off" && resolveRouter(config.playgroundFeatures && config.playgroundFeatures.pdDisagg, base) || null;
  const curlEnv = pdRouter && pdRouter.port != null ? {
    ...env,
    CURL_PORT: String(pdRouter.port)
  } : env;
  const curlText = interpolate(config.curl || "", curlEnv, modelName);
  const routerText = pdRouter && pdRouter.command ? interpolate(pdRouter.command, {
    ...env,
    PREFILL_PORT: PD_PORTS.prefill.serve,
    DECODE_PORT: PD_PORTS.decode.serve,
    ROUTER_PORT: pdRouter.port
  }, modelName) : "";
  const resetAll = () => setDeltas(initialDeltas());
  const placeholderGroups = (() => {
    const out = {
      command: [],
      curl: []
    };
    for (const [key, meta] of Object.entries(config.placeholders || ({}))) {
      (out[meta.target] || (out[meta.target] = [])).push({
        key,
        ...meta
      });
    }
    return out;
  })();
  const handleCopy = () => {
    navigator.clipboard.writeText(playgroundCommand);
    setCopied(true);
    setTimeout(() => setCopied(false), 1200);
  };
  const copyCurl = () => {
    navigator.clipboard.writeText(curlText);
    setCurlCopied(true);
    setTimeout(() => setCurlCopied(false), 1200);
  };
  const baseSummary = baseCell ? Object.entries(base).filter(([, v]) => v !== undefined && v !== "").map(([k, v]) => k === "hw" ? String(v).toUpperCase() : String(v)).join(" · ") : "(no verified cell at the current Deploy selection — showing playground only)";
  const renderChip = (label, current, value, onPick, opts = {}) => {
    const checked = current === value;
    const disabled = !!opts.disabled;
    return <span key={`${label}-${value === null ? "auto" : value}`} style={{
      ...s.chip,
      ...checked ? s.chipChecked : {},
      ...disabled ? s.chipDisabled : {}
    }} title={disabled ? opts.disabledReason || "Not available" : ""} onClick={() => {
      if (!disabled) onPick(value);
    }}>
        {label}
      </span>;
  };
  const renderSelect = (current, entries, onPick, base, labelFor, opts = {}) => {
    const hideSet = new Set(opts.hideValues || []);
    const items = [];
    for (const entry of entries || []) {
      const c = helpers.evaluateChip(entry, base);
      if (c.hidden) continue;
      if (hideSet.has(c.value)) continue;
      const lbl = labelFor ? labelFor(c) : c.label !== undefined ? c.label : c.value === null ? "Auto" : String(c.value);
      items.push({
        ...c,
        label: lbl
      });
    }
    let idx = items.findIndex(c => c.value === current);
    if (idx === -1) idx = 0;
    return <select style={{
      ...s.select,
      ...opts.disabled ? s.chipDisabled : {}
    }} disabled={!!opts.disabled} title={opts.disabled ? opts.disabledReason || "Not available" : ""} value={idx} onChange={e => {
      const next = items[parseInt(e.target.value, 10)];
      if (next && !next.disabled) onPick(next.value);
    }}>
        {items.map((c, i) => <option key={i} value={i} disabled={c.disabled}>
            {c.label}{c.disabled ? " (n/a)" : ""}
          </option>)}
      </select>;
  };
  return <div style={s.container} className="not-prose">
      {}
      <div style={s.baseStrip}>
        <span style={{
    fontWeight: 600
  }}>Inherited base from Deployment:</span>
        <code style={{
    fontFamily: "Menlo, monospace"
  }}>{baseSummary}</code>
        {}
        <button type="button" style={s.switchBaseBtn} onClick={() => {
    const el = document.getElementById(DEPLOYMENT_COMPONENT_ID) || document.getElementById("deployment") || document.getElementById("deploy");
    if (el) el.scrollIntoView({
      behavior: "smooth",
      block: "start"
    });
  }}>
          ↑ Switch base
        </button>
        <button style={s.resetBtn} onClick={resetAll}>Reset all overrides</button>
      </div>

      {}
      {Object.entries(AXIS_HANDLERS).map(([axisId, handler]) => {
    const fc = pgFeatures[axisId];
    if (!fc) return null;
    if (typeof fc.showWhen === "function" && !fc.showWhen(constraintBase)) return null;
    const setValue = next => setDeltas(d => ({
      ...d,
      [axisId]: next
    }));
    return handler.render({
      axisId,
      value: deltas[axisId],
      setValue,
      fc,
      base: constraintBase,
      s,
      h: helpers,
      renderChip,
      renderSelect,
      derived: derivedMap[axisId] || null
    });
  })}

      {}
      <div style={s.card}>
        <div style={s.title}>Playground Command (compare with base)</div>
        <div style={s.commandWrap}>
          <div style={s.commandHeader}>
            <div style={s.headerLeft}>
              <div style={s.badge(playgroundVerified)}>
                <span style={s.badgeDot(playgroundVerified)} />
                {playgroundVerified ? "Verified" : "Not Verified"}
              </div>
              {}
              {matchedSiblingCell && <span style={s.matchedHint}>
                  matches <code style={{
    fontFamily: "Menlo, monospace"
  }}>
                    {matchedSiblingCell.match.strategy}
                  </code>
                  <button type="button" style={s.matchedSwitchBtn} onClick={() => {
    const m = matchedSiblingCell.match;
    setDeltas(initialDeltas());
    const hash = new URLSearchParams(m).toString();
    window.location.hash = hash;
    window.dispatchEvent(new CustomEvent("sglang-deploy-sel", {
      detail: m
    }));
  }}>
                    switch base →
                  </button>
                </span>}
              <div style={s.runModeWrap} role="tablist" aria-label="Output format">
                <span style={s.runModeChip(runMode === "python")} onClick={() => setRunMode("python")} role="tab" aria-selected={runMode === "python"}>
                  Python
                </span>
                <span style={s.runModeChipLast(runMode === "docker")} onClick={() => setRunMode("docker")} role="tab" aria-selected={runMode === "docker"}>
                  Docker
                </span>
              </div>
            </div>
            <div style={s.iconRow}>
              <button style={s.iconButton} onClick={handleCopy}>
                {copied ? "✓ Copied" : "⧉ Copy"}
              </button>
              <button style={s.iconButton} onClick={() => setModal("curl")}>$ cURL</button>
              <button style={s.iconButton} onClick={() => setModal("env")}>⚙ Env</button>
              {}
              {!playgroundVerified && baseCell && <button style={{
    ...s.iconButton,
    borderColor: isDark ? "#FDBA74" : "#FB923C",
    color: isDark ? "#FDBA74" : "#C2410C",
    fontWeight: 600
  }} onClick={() => setModal("submit")} title="I verified this command on my hardware — open a pre-filled GitHub issue to land it as a cookbook cell.">
                  Submit ↗
                </button>}
            </div>
          </div>
          <pre style={s.commandPre}>
            {baseCell ? diffLines.map((d, i) => <span key={i} style={d.kind === "added" ? s.diffLineAdded : d.kind === "removed" ? s.diffLineRemoved : s.diffLineUnchanged}>
                {d.kind === "added" ? "+ " : d.kind === "removed" ? "- " : "  "}
                {d.line}{"\n"}
              </span>) : "# No verified base cell at the current Deployment selection.\n# Pick a supported hardware/variant in the Deployment panel to populate the playground base."}
          </pre>
          {pgSpecHint && <div style={s.mtpWarn}>
              {pgPdRole ? <>⚠️ Speculative decoding ({pgSpecAlgoName}) is on — for a target concurrency of N, set <code>--max-running-requests &lt;N*2&gt;</code> on <strong>both</strong> the prefill and decode roles. That ceiling is server-wide and floor-divided by <code>attn_dp_size</code>, so size the decode graphs to the per-rank batch it leaves: <code>--cuda-graph-bs-decode</code> up to <code>N*2 / dp_size</code>.</> : <>⚠️ Speculative decoding ({pgSpecAlgoName}) is on — SGLang resets <code>--max-running-requests</code> to <strong>48</strong> when it isn't set. Add <code>--max-running-requests &lt;N&gt;</code> sized for your target concurrency.</>}
            </div>}
          {pgCpDpHint && <div style={s.mtpWarn}>
              ⚠️ Interleave prefill-CP together with DP-Attention: current SGLang releases assert <code>dp_size == 1</code> for the interleave layout, so this command fails at startup. Combined CP + DP-Attention support is planned upstream — keep one of the two off until it lands.
            </div>}
        </div>
      </div>

      {}
      {pdRouter && routerText && <div style={s.card}>
          <div style={s.title}>Router</div>
          <div style={{
    fontSize: 11,
    opacity: 0.7,
    margin: "0 0 6px"
  }}>
            Run after both roles are up. Substitute <code>{"<prefill-host>"}</code> /{" "}
            <code>{"<decode-host>"}</code> with reachable hosts (both <code>127.0.0.1</code>{" "}
            on a same-host deployment). Client traffic (cURL) targets this router.
          </div>
          <div style={s.commandWrap}>
            <div style={s.commandHeader}>
              <div style={{
    fontSize: 11,
    opacity: 0.7
  }}>port {pdRouter.port}</div>
              <button style={s.iconButton} onClick={() => {
    navigator.clipboard.writeText(routerText);
    setRouterCopied(true);
    setTimeout(() => setRouterCopied(false), 1200);
  }}>
                {routerCopied ? "✓ Copied" : "⧉ Copy"}
              </button>
            </div>
            <pre style={s.commandPre}>{routerText}</pre>
          </div>
        </div>}

      {}
      {modal === "curl" && <dialog ref={openDialog} style={s.dialog} onClose={() => setModal(null)} onClick={onDialogClick}>
          <div style={s.modalHeader}>
            <div style={s.modalTitle}>cURL example</div>
            <button style={s.modalCloseBtn} onClick={() => setModal(null)} aria-label="Close">×</button>
          </div>
          <div style={s.commandWrap}>
            <div style={s.commandHeader}>
              <div style={{
    fontSize: 11,
    opacity: 0.7
  }}>
                Model: <code>{modelName || "(unresolved)"}</code>
              </div>
              <button style={s.iconButton} onClick={copyCurl}>
                {curlCopied ? "✓ Copied" : "⧉ Copy"}
              </button>
            </div>
            <pre style={s.commandPre}>{curlText}</pre>
          </div>
          {pdRouter && <p style={{
    fontSize: 11,
    opacity: 0.85,
    marginTop: 8
  }}>
              <strong>PD-Disaggregation active</strong> — this targets the router on
              {" "}<code>:{pdRouter.port}</code>; client traffic must not hit the role
              servers directly.
            </p>}
          <p style={{
    fontSize: 11,
    opacity: 0.7,
    marginTop: 8
  }}>
            Edit <code>CURL_HOST</code> / <code>CURL_PORT</code> in the Env panel.
          </p>
        </dialog>}

      {}
      {modal === "env" && <dialog ref={openDialog} style={s.dialog} onClose={() => setModal(null)} onClick={onDialogClick}>
          <div style={s.modalHeader}>
            <div style={s.modalTitle}>Env / placeholder values</div>
            <button style={s.modalCloseBtn} onClick={() => setModal(null)} aria-label="Close">×</button>
          </div>
            {placeholderGroups.curl.length > 0 && <div>
                <div style={s.sectionHeading}>cURL placeholders</div>
                {placeholderGroups.curl.map(({key, label}) => <div key={key} style={s.formField}>
                    <label style={s.formLabel}>
                      {label} <code style={{
    opacity: 0.6
  }}>{`{{${key}}}`}</code>
                    </label>
                    <input style={s.formInput} value={envDraft[key] ?? ""} onChange={e => setEnvDraft({
    ...envDraft,
    [key]: e.target.value
  })} />
                  </div>)}
              </div>}
            {placeholderGroups.command.length > 0 && <div>
                <div style={s.sectionHeading}>Command placeholders</div>
                {placeholderGroups.command.map(({key, label}) => <div key={key} style={s.formField}>
                    <label style={s.formLabel}>
                      {label} <code style={{
    opacity: 0.6
  }}>{`{{${key}}}`}</code>
                    </label>
                    <input style={s.formInput} value={envDraft[key] ?? ""} onChange={e => setEnvDraft({
    ...envDraft,
    [key]: e.target.value
  })} />
                  </div>)}
              </div>}
            <div style={{
    display: "flex",
    justifyContent: "flex-end",
    gap: 8,
    marginTop: 16
  }}>
              <button style={{
    ...s.iconButton,
    padding: "6px 14px"
  }} onClick={() => setModal(null)}>Cancel</button>
              <button style={s.primaryBtn} onClick={() => {
    saveEnv(envDraft);
    setModal(null);
  }}>Save</button>
            </div>
          <p style={{
    fontSize: 11,
    opacity: 0.7,
    marginTop: 10
  }}>
            Values persist in localStorage and are shared with the Deployment panel.
          </p>
        </dialog>}

      {}
      {modal === "submit" && <dialog ref={openDialog} style={s.dialog} onClose={() => setModal(null)} onClick={onDialogClick}>
            <div style={s.modalHeader}>
              <div style={s.modalTitle}>Submit verified cell</div>
              <button style={s.modalCloseBtn} onClick={() => setModal(null)} aria-label="Close">×</button>
            </div>
            <p style={{
    fontSize: 12,
    opacity: 0.85,
    marginTop: 0,
    marginBottom: 12
  }}>
              You've put together a combination that isn't in the verified
              catalog yet. After you've run the command end-to-end on the
              target hardware, this submits a pre-filled GitHub Issue that a
              maintainer can convert into a PR.
            </p>

            <div style={s.sectionHeading}>Combination</div>
            <code style={{
    fontFamily: "Menlo, monospace",
    fontSize: 12
  }}>
              {base.hw} / {base.variant} / {base.quant} / {base.strategy} / {base.nodes}
            </code>
            {}
            {(() => {
    const adds = diffLines.filter(d => d.kind === "added");
    const rems = diffLines.filter(d => d.kind === "removed");
    if (adds.length === 0 && rems.length === 0) return null;
    return <>
                  <div style={{
      ...s.sectionHeading,
      marginTop: 10
    }}>
                    Overrides vs base ({adds.length} added · {rems.length} removed)
                  </div>
                  <pre style={{
      margin: 0,
      padding: "8px 10px",
      background: isDark ? "#111827" : "#f5f5f5",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      borderRadius: 4,
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace",
      fontSize: 12,
      lineHeight: 1.4,
      maxHeight: 160,
      overflowY: "auto",
      whiteSpace: "pre-wrap"
    }}>
                    {[...rems, ...adds].map((d, i) => <div key={i} style={d.kind === "added" ? s.diffLineAdded : s.diffLineRemoved}>
                        {d.kind === "added" ? "+ " : "- "}
                        {d.line.replace(/^\s*/, "")}
                      </div>)}
                  </pre>
                </>;
  })()}

            <div style={{
    ...s.sectionHeading,
    marginTop: 14
  }}>Attestation (all required)</div>
            <div style={s.formField}>
              <label style={{
    fontSize: 12,
    display: "flex",
    alignItems: "flex-start",
    gap: 6
  }}>
                <input type="checkbox" checked={submitAttest.ranCommand} onChange={e => setSubmitAttest({
    ...submitAttest,
    ranCommand: e.target.checked
  })} />
                I ran this exact command on the listed hardware.
              </label>
              <label style={{
    fontSize: 12,
    display: "flex",
    alignItems: "flex-start",
    gap: 6
  }}>
                <input type="checkbox" checked={submitAttest.reachedReady} onChange={e => setSubmitAttest({
    ...submitAttest,
    reachedReady: e.target.checked
  })} />
                The server reached READY and answered a cURL request successfully.
              </label>
              <label style={{
    fontSize: 12,
    display: "flex",
    alignItems: "flex-start",
    gap: 6
  }}>
                <input type="checkbox" checked={submitAttest.outputCorrect} onChange={e => setSubmitAttest({
    ...submitAttest,
    outputCorrect: e.target.checked
  })} />
                Output looked correct on at least one prompt.
              </label>
            </div>

            <div style={{
    ...s.sectionHeading,
    marginTop: 14
  }}>SGLang version (required)</div>
            <input style={{
    ...s.formInput,
    width: "100%",
    boxSizing: "border-box"
  }} placeholder="sglang==0.5.4  (or git SHA abc1234)" value={submitDraft.sglangVersion} onChange={e => setSubmitDraft({
    ...submitDraft,
    sglangVersion: e.target.value
  })} />

            <div style={{
    ...s.sectionHeading,
    marginTop: 14
  }}>Benchmark result (optional)</div>
            <input style={{
    ...s.formInput,
    width: "100%",
    boxSizing: "border-box"
  }} placeholder="TTFT 95 ms / TPOT 18 ms / 1820 tok/s @ bs=64" value={submitDraft.benchResult} onChange={e => setSubmitDraft({
    ...submitDraft,
    benchResult: e.target.value
  })} />

            <div style={{
    ...s.sectionHeading,
    marginTop: 14
  }}>Notes / caveats (optional)</div>
            <textarea style={{
    ...s.formInput,
    width: "100%",
    boxSizing: "border-box",
    minHeight: 110,
    resize: "vertical",
    fontFamily: "inherit"
  }} placeholder="Cluster config, env-var quirks, NIC mappings, multi-node bootstrap details, …" value={submitDraft.notes} onChange={e => setSubmitDraft({
    ...submitDraft,
    notes: e.target.value
  })} />

            <div style={{
    display: "flex",
    justifyContent: "flex-end",
    gap: 8,
    marginTop: 16,
    alignItems: "center"
  }}>
              {!submitReady && <span style={{
    fontSize: 11,
    opacity: 0.7,
    marginRight: "auto"
  }}>
                  Tick all attestations and fill SGLang version to enable submit.
                </span>}
              <button style={{
    ...s.iconButton,
    padding: "6px 14px"
  }} onClick={() => setModal(null)}>Cancel</button>
              <a href={submitReady ? submitUrl : undefined} target="_blank" rel="noopener noreferrer" onClick={e => {
    if (!submitReady) e.preventDefault(); else setModal(null);
  }} style={{
    ...s.primaryBtn,
    textDecoration: "none",
    display: "inline-flex",
    alignItems: "center",
    opacity: submitReady ? 1 : 0.4,
    cursor: submitReady ? "pointer" : "not-allowed"
  }}>
                Open submission on GitHub →
              </a>
            </div>
          <p style={{
    fontSize: 11,
    opacity: 0.7,
    marginTop: 10
  }}>
            The CTA opens a pre-filled GitHub Issue using the
            <code> 3-playground-verified-cell.yml</code> template. A
            maintainer with the listed hardware will review and convert it
            into a cookbook PR.
          </p>
        </dialog>}
    </div>;
};

export const benchmarks = [{
  match: {
    hw: "gb300",
    strategy: "low-latency"
  },
  sglang_version: "d6ab04bdf1",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 1,
      num_prompts: 8
    },
    ttft_ms: 219.15,
    tpot_ms: 3.94,
    tokens_per_sec_per_gpu: 261.03
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 598.41,
    tpot_ms: 6.43,
    tokens_per_sec_per_gpu: 2280.83
  }],
  accuracy: {
    gsm8k_pct: 97.50
  },
  notes: "Measured on 4x GB300 (TP4/EP4) with the final weights (zai-org/GLM-5.3-Flash, c5b82b63e37b) on the current release-image tree (d6ab04bdf1), adaptive MTP 5/1/6 with SGLANG_SIMULATE_ACC_LEN=3 (accept length confirmed 3.00 in the bench summary and server log): 80 random requests at 1,024 input / 256 output tokens and concurrency 16 produced 1,824.66 aggregate output tok/s after two discarded warmups. Simulated accept length makes this a throughput-mechanism number. Re-verified on the current release image (tree fe236ea6c3) within 1.5% on 4x GB300; the concurrency-1 entry (8 requests) comes from that re-run. Accuracy is from the shared non-simulated full GSM8K gate: 97.50% with a 100% stop rate over all 1,319 problems."
}, {
  match: {
    hw: "gb300",
    strategy: "low-latency",
    kvDsaPair: "fp8-trtllm",
    quant: "fp8"
  },
  sglang_version: "b3dc0388ed",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1000,
      osl: 1000,
      max_concurrency: 1,
      num_prompts: 80
    },
    ttft_ms: 204.79,
    tpot_ms: 2.0,
    tokens_per_sec_per_gpu: 227.06
  }, {
    workload: {
      dataset: "random",
      isl: 1000,
      osl: 1000,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 367.48,
    tpot_ms: 4.8,
    tokens_per_sec_per_gpu: 1492.93
  }, {
    workload: {
      dataset: "random",
      isl: 1000,
      osl: 1000,
      max_concurrency: 80,
      num_prompts: 80
    },
    ttft_ms: 1243.26,
    tpot_ms: 6.28,
    tokens_per_sec_per_gpu: 4487.92
  }, {
    workload: {
      dataset: "random",
      isl: 8000,
      osl: 1000,
      max_concurrency: 1,
      num_prompts: 80
    },
    ttft_ms: 237.02,
    tpot_ms: 1.95,
    tokens_per_sec_per_gpu: 1029.64
  }, {
    workload: {
      dataset: "random",
      isl: 8000,
      osl: 1000,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 986.68,
    tpot_ms: 5.16,
    tokens_per_sec_per_gpu: 5550.31
  }, {
    workload: {
      dataset: "random",
      isl: 8000,
      osl: 1000,
      max_concurrency: 80,
      num_prompts: 80
    },
    ttft_ms: 7085.4,
    tpot_ms: 11.65,
    tokens_per_sec_per_gpu: 8557.31
  }],
  notes: "The published Low Latency command with FP8 KV + TRT-LLM DSA (TP4, EP1, flashinfer_trtllm MoE, fixed MTP 5/1/6 with natural acceptance) on 4x GB300, zai-org/GLM-5.3-Flash (eb9eb208eb0d) on upstream main b3dc0388ed with FlashInfer 0.6.18, measured in the 2026-09 GLM-5.3-Flash perf campaign (run 20260911_glm53flash_vs_vllm, lane sglang-mtp-trtllm-ep1). Each row is 80 ShareGPT-length random-token requests over streaming /v1/completions with ignore_eos, temperature 0, seed 42, prefix cache disabled, --context-length 16384, --max-running-requests 80, --cuda-graph-max-bs-decode 80 and --mem-fraction-static 0.90; chat is 1,000 input / 1,000 output tokens and summarization 8,000 / 1,000. TTFT and TPOT are per-run means; c16 and c80 are the median of two runs, c1 a single run. Aggregate output tok/s: chat 454.11 / 2,985.85 / 8,975.84 and summarization 457.62 / 2,466.80 / 3,803.25 at concurrency 1 / 16 / 80, at accept lengths of 5.98 / 5.98 / 4.7 (chat) and 6.0 / 6.0 / 4.5 (summarization). In the same campaign, fixed-depth MTP drafting measured 8,490.07 chat c80 tok/s against 3,664.53 for adaptive drafting on the TP4/EP4 deep_gemm command, and EP1 + flashinfer_trtllm then improved every cell over that fixed-depth command by 4.4-13.6%. The campaign's GSM8K natural-stop gate (all 1,319 problems, reasoning effort low, 8,192 max tokens, temperature 1.0 / top_p 0.95) ran these launch flags on eleven source candidates and every run stopped naturally on all 1,319 problems at 90.8-92.7%; that protocol differs from the thinking-mode GSM8K behind the accuracy figures on this page."
}, {
  match: {
    hw: "gb300",
    strategy: "low-latency",
    kvDsaPair: "fp8-trtllm",
    dcp: "4",
    quant: "fp8"
  },
  sglang_version: "d6ab04bdf1",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 534.1,
    tpot_ms: 7.31,
    tokens_per_sec_per_gpu: 2100.76
  }],
  notes: "The Low Latency recipe with FP8 KV + TRT-LLM DSA and DCP4 (--dcp-size 4 --dcp-comm-backend a2a --dcp-replicate-q-proj) on 4x GB300, final weights (c5b82b63e37b) on the d6ab04bdf1 tree, adaptive MTP 5/1/6 with full decode graph: 80 random requests at 1,024 input / 256 output tokens and concurrency 16 produced 1,680.61 aggregate output tok/s at a 3.937 accept length — about 10% below the non-DCP FP8 Low Latency row. TRT-LLM DSA DCP decode returns the LSE natively, so this arm needs no patch. Re-verified on the current release image (tree fe236ea6c3) within 1.3% on 4x GB300."
}, {
  match: {
    hw: "gb300",
    strategy: "low-latency",
    kvDsaPair: "bf16-tilelang",
    dcp: "4",
    quant: "fp8"
  },
  sglang_version: "d6ab04bdf1",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 510.0,
    tpot_ms: 8.0,
    tokens_per_sec_per_gpu: 1957.25
  }],
  notes: "The Low Latency recipe with BF16 KV + TileLang DSA and DCP4 on 4x GB300, final weights (c5b82b63e37b) on the d6ab04bdf1 tree, adaptive MTP 5/1/6 with full decode graph: 80 random requests at 1,024 input / 256 output tokens and concurrency 16 produced 1,565.8 aggregate output tok/s at a 3.90 accept length. TileLang DSA DCP decode needs the LSE fix that ships in the current release image. Re-verified on the current release image (tree fe236ea6c3) within 1.3% on 4x GB300."
}, {
  match: {
    hw: "gb300",
    strategy: "high-throughput"
  },
  sglang_version: "d6ab04bdf1",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 586.33,
    tpot_ms: 11.51,
    tokens_per_sec_per_gpu: 1451.53
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 64,
      num_prompts: 320
    },
    ttft_ms: 1461.23,
    tpot_ms: 18.36,
    tokens_per_sec_per_gpu: 3325.3
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 256,
      num_prompts: 1280
    },
    ttft_ms: 3912.74,
    tpot_ms: 37.69,
    tokens_per_sec_per_gpu: 6035.41
  }],
  accuracy: {
    gsm8k_pct: 97.50
  },
  notes: "Measured on 4x GB300 (TP4/EP4) with the final weights (zai-org/GLM-5.3-Flash, c5b82b63e37b) on the current release-image tree (d6ab04bdf1), speculative decoding off, after two discarded warmups per row: 1,161.22 / 2,660.24 / 4,828.33 aggregate output tok/s at concurrency 16 / 64 / 256 (80 / 320 / 1,280 random requests at 1,024 input / 256 output tokens). The server ran exactly the published cell command. Throughput at 256 is still scaling but sublinear (prefill queueing). Accuracy is from the shared non-simulated full GSM8K gate: 97.50% with a 100% stop rate over all 1,319 problems. With HiCache L1+L2 (32 GB host tier, 16k prefill chunks) the same protocol measured 1,202.07 / 2,696.20 / 4,634.47 tok/s — within 4% of the non-HiCache rows; the random dataset has no prefix reuse, so L2 benefit was not exercised. Re-verified on the current release image (tree fe236ea6c3) within 1.5% on 4x GB300."
}, {
  match: {
    hw: "gb300",
    strategy: "high-throughput",
    kvDsaPair: "fp8-trtllm",
    quant: "fp8"
  },
  sglang_version: "d6ab04bdf1",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 581.52,
    tpot_ms: 10.79,
    tokens_per_sec_per_gpu: 1533.84
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 64,
      num_prompts: 320
    },
    ttft_ms: 1459.62,
    tpot_ms: 17.68,
    tokens_per_sec_per_gpu: 3423.26
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 256,
      num_prompts: 1280
    },
    ttft_ms: 3908.3,
    tpot_ms: 36.12,
    tokens_per_sec_per_gpu: 6221.28
  }],
  accuracy: {
    gsm8k_pct: 97.35
  },
  notes: "FP8 KV cache with TRT-LLM DSA on 4x GB300, final weights (c5b82b63e37b) on the current release-image tree (d6ab04bdf1), same protocol as the BF16 rows: 1,227.07 / 2,738.61 / 4,977.02 aggregate output tok/s at concurrency 16 / 64 / 256 — 2.9–5.7% above BF16 + TileLang across the curve, and the FP8 pool holds 12.6M tokens per rank vs 7.0M at BF16 (1.8x capacity at identical pool bytes). Accuracy is the full GSM8K gate on this variant: 97.35% vs 97.50% on BF16 KV, a 0.15-point gap inside sampling noise, with a 100% stop rate over all 1,319 problems. With HiCache L1+L2 (32 GB host tier, 16k prefill chunks) the same protocol measured 1,263.85 / 2,763.31 / 4,773.95 tok/s — within 5% of the non-HiCache rows; the random dataset has no prefix reuse, so L2 benefit was not exercised. Re-verified on the current release image (tree fe236ea6c3) within 1.5% on 4x GB300."
}, {
  match: {
    hw: "gb300",
    strategy: "low-latency",
    quant: "nvfp4",
    kvDsaPair: "bf16-tilelang"
  },
  sglang_version: "fe236ea6c3",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 1,
      num_prompts: 8
    },
    ttft_ms: 175.18,
    tpot_ms: 3.84,
    tokens_per_sec_per_gpu: 276.74
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 479.9,
    tpot_ms: 6.33,
    tokens_per_sec_per_gpu: 2437.28
  }],
  accuracy: {
    gsm8k_pct: 97.14,
    aime2026_pct: 92.45
  },
  notes: "RadixArk/GLM-5.3-Flash-NVFP4 — NVFP4 W4A4 post-training quantization of zai-org/GLM-5.3-Flash-BF16 with NVIDIA Model Optimizer 0.46.0 (abs-max scaling, group size 16): routed and shared experts plus the dense MLPs are FP4, while all attention (KDA, DSA indexer, MLA), the router, norms, the vision tower, the MTP layer, embeddings, and the LM head stay BF16. Accuracy measured on 4x GB300 with the lmsysorg/sglang:glm-5.3-flash image (PR #36507 head 033446bb05), adaptive MTP 5/1/6, BF16 KV + TileLang DSA, and the flashinfer_cutlass MoE runner. GSM8K 97.14% over the full 1,319-example split x 4 seeds (per-seed range 96.89-97.42%, stop rate 99.85-100%) and AIME 2026 92.45% (30 problems x 16 repeats x 4 seeds = 1,920 generations, per-seed range 91.67-93.54%), both at temperature 1.0 / top_p 0.95. The accuracy runs used the NEXTN spelling of --speculative-algorithm, which resolves to the same runtime path as the published EAGLE command on this tree. Speed measured on 4x GB300 with the current release image (tree fe236ea6c3) with SGLANG_SIMULATE_ACC_LEN=3 pinning the accept length (3.00): 8 requests at concurrency 1 and 80 at concurrency 16 (1,024 input / 256 output tokens) produced 221.39 and 1,949.82 aggregate output tok/s after two discarded warmups — the c16 cell lands within 5.3% of the B300 measurement, while the c1 cell runs ~20% lower on the Grace host (a host-latency effect visible only at bs1). Simulated accept length makes these throughput-mechanism numbers."
}, {
  match: {
    hw: "gb300",
    strategy: "high-throughput",
    quant: "nvfp4",
    kvDsaPair: "bf16-tilelang"
  },
  sglang_version: "fe236ea6c3",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 447.92,
    tpot_ms: 10.09,
    tokens_per_sec_per_gpu: 1691.08
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 64,
      num_prompts: 320
    },
    ttft_ms: 1094.05,
    tpot_ms: 15.19,
    tokens_per_sec_per_gpu: 4114.2
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 256,
      num_prompts: 1280
    },
    ttft_ms: 2890.32,
    tpot_ms: 31.85,
    tokens_per_sec_per_gpu: 7399.36
  }],
  accuracy: {
    gsm8k_pct: 97.14,
    aime2026_pct: 92.45
  },
  notes: "RadixArk/GLM-5.3-Flash-NVFP4 with speculative decoding off — same checkpoint, image, and 4x GB300 measurement stack as the NVFP4 Low Latency row (ModelOpt 0.46.0 NVFP4 W4A4, abs-max, group size 16; MoE and dense MLPs in FP4, attention/router/MTP/embeddings BF16). Accuracy is a checkpoint-level result carried from that arm: GSM8K 97.14% over the full 1,319-example split x 4 seeds (per-seed range 96.89-97.42%, stop rate 99.85-100%) and AIME 2026 92.45% (30 problems x 16 repeats x 4 seeds, per-seed range 91.67-93.54%). Those runs used the NEXTN spelling of --speculative-algorithm on the adaptive-MTP arm, which resolves to the same runtime path as the published EAGLE command. Speed measured on 4x GB300 with the current release image (tree fe236ea6c3): 80 / 320 / 1,280 random requests at concurrency 16 / 64 / 256 (1,024 input / 256 output tokens) produced 1,352.86 / 3,291.36 / 5,919.49 aggregate output tok/s after two discarded warmups — at or above B300 parity, and above the FP8 gb300 High Throughput row at c256 (5,919.49 vs 4,828.33), as expected for W4A4."
}, {
  match: {
    hw: "gb300",
    strategy: "low-latency",
    quant: "nvfp4",
    kvDsaPair: "fp8-trtllm"
  },
  sglang_version: "fe236ea6c3",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 1,
      num_prompts: 8
    },
    ttft_ms: 176.24,
    tpot_ms: 3.6,
    tokens_per_sec_per_gpu: 291.64
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 475.37,
    tpot_ms: 6.09,
    tokens_per_sec_per_gpu: 2517.1
  }],
  notes: "FP8 KV + TRT-LLM DSA pairing of the NVFP4 recipe, TP4-only with the flashinfer_cutlass MoE runner and adaptive MTP 5/1/6. Speed measured on 4x GB300 with the current release image (tree fe236ea6c3) with SGLANG_SIMULATE_ACC_LEN=3 pinning the accept length (3.00): 8 requests at concurrency 1 and 80 at concurrency 16 (1,024 input / 256 output tokens) produced 233.31 and 2,013.68 aggregate output tok/s after two discarded warmups — the c16 cell lands within 3.4% of the B300 measurement, while the c1 cell runs ~20% lower on the Grace host (a host-latency effect visible only at bs1). Simulated accept length makes these throughput-mechanism numbers."
}, {
  match: {
    hw: "gb300",
    strategy: "high-throughput",
    quant: "nvfp4",
    kvDsaPair: "fp8-trtllm"
  },
  sglang_version: "fe236ea6c3",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 16,
      num_prompts: 80
    },
    ttft_ms: 447.34,
    tpot_ms: 9.38,
    tokens_per_sec_per_gpu: 1799.06
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 64,
      num_prompts: 320
    },
    ttft_ms: 1090.95,
    tpot_ms: 14.4,
    tokens_per_sec_per_gpu: 4285.73
  }, {
    workload: {
      dataset: "random",
      isl: 1024,
      osl: 256,
      max_concurrency: 256,
      num_prompts: 1280
    },
    ttft_ms: 2891.78,
    tpot_ms: 30.2,
    tokens_per_sec_per_gpu: 7688.08
  }],
  notes: "FP8 KV + TRT-LLM DSA pairing of the NVFP4 recipe with speculative decoding off — same TP4-only flashinfer_cutlass stack as the NVFP4 fp8-trtllm Low Latency row. Speed measured on 4x GB300 with the current release image (tree fe236ea6c3): 80 / 320 / 1,280 random requests at concurrency 16 / 64 / 256 (1,024 input / 256 output tokens) produced 1,439.25 / 3,428.58 / 6,150.46 aggregate output tok/s after two discarded warmups — at or above B300 parity, and above the FP8 gb300 High Throughput row at c256 (6,150.46 vs 4,977.02), as expected for W4A4."
}, {
  match: {
    hw: "h100",
    strategy: "low-latency"
  },
  sglang_version: "f040cc72e6",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 8192,
      osl: 1024,
      max_concurrency: 1,
      num_prompts: 8
    },
    ttft_ms: 345.66,
    tpot_ms: 4.37,
    tokens_per_sec_per_gpu: 239.19
  }, {
    workload: {
      dataset: "random",
      isl: 8192,
      osl: 1024,
      max_concurrency: 16,
      num_prompts: 32
    },
    ttft_ms: 3442.21,
    tpot_ms: 10.24,
    tokens_per_sec_per_gpu: 1323.44
  }],
  accuracy: {
    gsm8k_pct: 97.27
  },
  notes: "Full GSM8K (all 1,319 problems) on 8x H100 (TP8/EP8) with zai-org/GLM-5.3-Flash at f040cc72e6: 97.27%. Run with `sgl-eval run gsm8k --base-url http://localhost:30000/v1 --num-threads 32 --max-tokens 32768`; gsm8k's registered default leaves thinking off, so these are non-thinking numbers and are not directly comparable to the GB300 rows above. Speed was measured by the Z.ai team on the published image lmsysorg/sglang:glm-5.3-flash (sha256:aa9210e3…, tree fe236ea6c3), on the published Low Latency cell recipe with the BF16 KV + TileLang DSA pairing and SGLANG_SIMULATE_ACC_LEN=3 in the environment (accept 3.00 at c1, 2.98 at c16): random 8,192-input / 1,024-output requests (range ratio 1, sglang-oai backend) produced 212.61 aggregate output tok/s at concurrency 1 (8 requests) and 1,176.39 at concurrency 16 (32 requests) after two discarded warmups — simulated accept length makes these throughput-mechanism numbers. With HiCache L1+L2 the same protocol measured 1,206.11 at c16 (TTFT 2,985.08 ms, TPOT 10.35 ms). The retokenized-token count in these runs is about a quarter of the generated tokens because reasoning_content is not retokenized (benign)."
}, {
  match: {
    hw: "h100",
    strategy: "high-throughput"
  },
  sglang_version: "f040cc72e6",
  accuracy: {
    gsm8k_pct: 97.50
  },
  notes: "Full GSM8K (all 1,319 problems) on 8x H100 (TP8/EP8) with zai-org/GLM-5.3-Flash at f040cc72e6: 97.50% for the recommended selection; 97.27-97.50% across all 4 measured selections. Run with `sgl-eval run gsm8k --base-url http://localhost:30000/v1 --num-threads 32 --max-tokens 32768`; gsm8k's registered default leaves thinking off, so these are non-thinking numbers and are not directly comparable to the GB300 rows above. Accuracy only, no speed measurement."
}, {
  match: {
    hw: "h200",
    strategy: "low-latency"
  },
  sglang_version: "f040cc72e6",
  accuracy: {
    gsm8k_pct: 97.04
  },
  notes: "Full GSM8K (all 1,319 problems) on 8x H200 (TP8/EP8) with zai-org/GLM-5.3-Flash at f040cc72e6: 97.04%. Run with `sgl-eval run gsm8k --base-url http://localhost:30000/v1 --num-threads 32 --max-tokens 32768`; gsm8k's registered default leaves thinking off, so these are non-thinking numbers and are not directly comparable to the GB300 rows above. Accuracy only, no speed measurement."
}, {
  match: {
    hw: "h200",
    strategy: "high-throughput"
  },
  sglang_version: "f040cc72e6",
  accuracy: {
    gsm8k_pct: 97.35
  },
  notes: "Full GSM8K (all 1,319 problems) on 8x H200 (TP8/EP8) with zai-org/GLM-5.3-Flash at f040cc72e6: 97.35% for the recommended selection; 97.19-97.57% across all 4 measured selections. Run with `sgl-eval run gsm8k --base-url http://localhost:30000/v1 --num-threads 32 --max-tokens 32768`; gsm8k's registered default leaves thinking off, so these are non-thinking numbers and are not directly comparable to the GB300 rows above. Accuracy only, no speed measurement."
}, {
  match: {
    hw: "b200",
    strategy: "low-latency"
  },
  sglang_version: "f040cc72e6",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 8192,
      osl: 1024,
      max_concurrency: 1,
      num_prompts: 8
    },
    ttft_ms: 352.45,
    tpot_ms: 3.75,
    tokens_per_sec_per_gpu: 274.8
  }, {
    workload: {
      dataset: "random",
      isl: 8192,
      osl: 1024,
      max_concurrency: 16,
      num_prompts: 32
    },
    ttft_ms: 2456.16,
    tpot_ms: 6.43,
    tokens_per_sec_per_gpu: 2037.52
  }],
  accuracy: {
    gsm8k_pct: 97.27
  },
  notes: "Full GSM8K (all 1,319 problems) on 8x B200 (TP8/EP8) with zai-org/GLM-5.3-Flash at f040cc72e6: 97.27% for the recommended selection; 97.12-97.27% across all 4 measured selections. Run with `sgl-eval run gsm8k --base-url http://localhost:30000/v1 --num-threads 32 --max-tokens 32768`; gsm8k's registered default leaves thinking off, so these are non-thinking numbers and are not directly comparable to the GB300 rows above. Speed was measured by the Z.ai team on the published image lmsysorg/sglang:glm-5.3-flash (sha256:aa9210e3…, tree fe236ea6c3), on the published Low Latency cell recipe with the BF16 KV + TileLang DSA pairing and SGLANG_SIMULATE_ACC_LEN=3 in the environment (accept 3.00 at c1, 2.98 at c16): random 8,192-input / 1,024-output requests (range ratio 1, sglang-oai backend) produced 244.27 aggregate output tok/s at concurrency 1 (8 requests) and 1,811.13 at concurrency 16 (32 requests) after two discarded warmups — simulated accept length makes these throughput-mechanism numbers; the FP8 KV + TRT-LLM pairing recommended on Blackwell was not speed-measured on this platform. With HiCache L1+L2 the same protocol measured 258.05 at c1 (TTFT 245.25 ms) and 2,136.62 at c16 (TTFT 1,590.40 ms)."
}, {
  match: {
    hw: "b200",
    strategy: "high-throughput"
  },
  sglang_version: "f040cc72e6",
  accuracy: {
    gsm8k_pct: 97.27
  },
  notes: "Full GSM8K (all 1,319 problems) on 8x B200 (TP8/EP8) with zai-org/GLM-5.3-Flash at f040cc72e6: 97.27% for the recommended selection; 96.97-97.35% across all 8 measured selections. Run with `sgl-eval run gsm8k --base-url http://localhost:30000/v1 --num-threads 32 --max-tokens 32768`; gsm8k's registered default leaves thinking off, so these are non-thinking numbers and are not directly comparable to the GB300 rows above. Accuracy only, no speed measurement."
}, {
  match: {
    hw: "b300",
    strategy: "low-latency"
  },
  sglang_version: "f040cc72e6",
  latencyPercentile: "Mean",
  speed: [{
    workload: {
      dataset: "random",
      isl: 8192,
      osl: 1024,
      max_concurrency: 1,
      num_prompts: 8
    },
    ttft_ms: 230.34,
    tpot_ms: 3.37,
    tokens_per_sec_per_gpu: 313.02
  }, {
    workload: {
      dataset: "random",
      isl: 8192,
      osl: 1024,
      max_concurrency: 16,
      num_prompts: 32
    },
    ttft_ms: 1908.02,
    tpot_ms: 5.83,
    tokens_per_sec_per_gpu: 2338.81
  }],
  accuracy: {
    gsm8k_pct: 96.82
  },
  notes: "Full GSM8K (all 1,319 problems) on 8x B300 (TP8/EP8) with zai-org/GLM-5.3-Flash at f040cc72e6: 96.82% for the recommended selection; 96.82-97.27% across all 4 measured selections. Run with `sgl-eval run gsm8k --base-url http://localhost:30000/v1 --num-threads 32 --max-tokens 32768`; gsm8k's registered default leaves thinking off, so these are non-thinking numbers and are not directly comparable to the GB300 rows above. Speed was measured by the Z.ai team on the published image lmsysorg/sglang:glm-5.3-flash (sha256:aa9210e3…, tree fe236ea6c3), on the published Low Latency cell recipe with the BF16 KV + TileLang DSA pairing and SGLANG_SIMULATE_ACC_LEN=3 in the environment (accept 3.00 at c1, 2.98 at c16): random 8,192-input / 1,024-output requests (range ratio 1, sglang-oai backend) produced 278.24 aggregate output tok/s at concurrency 1 (8 requests) and 2,078.94 at concurrency 16 (32 requests) after two discarded warmups — simulated accept length makes these throughput-mechanism numbers; the FP8 KV + TRT-LLM pairing recommended on Blackwell was not speed-measured on this platform. With HiCache L1+L2 the same protocol measured 278.01 at c1 (TTFT 214.12 ms) and 2,323.89 at c16 (TTFT 1,433.37 ms)."
}, {
  match: {
    hw: "b300",
    strategy: "high-throughput"
  },
  sglang_version: "f040cc72e6",
  accuracy: {
    gsm8k_pct: 96.97
  },
  notes: "Full GSM8K (all 1,319 problems) on 8x B300 (TP8/EP8) with zai-org/GLM-5.3-Flash at f040cc72e6: 96.97% for the recommended selection; 96.97-97.04% across all 8 measured selections. Run with `sgl-eval run gsm8k --base-url http://localhost:30000/v1 --num-threads 32 --max-tokens 32768`; gsm8k's registered default leaves thinking off, so these are non-thinking numbers and are not directly comparable to the GB300 rows above. Accuracy only, no speed measurement."
}, {
  match: {
    hw: "gb200",
    strategy: "low-latency"
  }
}, {
  match: {
    hw: "gb200",
    strategy: "high-throughput"
  }
}, {
  match: {
    hw: "gb200",
    strategy: "low-latency",
    quant: "nvfp4"
  }
}, {
  match: {
    hw: "gb200",
    strategy: "high-throughput",
    quant: "nvfp4"
  }
}, {
  match: {
    hw: "b200",
    strategy: "low-latency",
    quant: "nvfp4"
  }
}, {
  match: {
    hw: "b200",
    strategy: "high-throughput",
    quant: "nvfp4"
  }
}, {
  match: {
    hw: "b300",
    strategy: "low-latency",
    quant: "nvfp4"
  }
}, {
  match: {
    hw: "b300",
    strategy: "high-throughput",
    quant: "nvfp4"
  }
}];

export const config = {
  modelName: "GLM-5.3-Flash",
  supportedHardware: ["gb300", "h100", "h200", "b200", "b300", "gb200"],
  matchDims: [{
    id: "strategy",
    title: "Strategy",
    options: [{
      id: "low-latency",
      label: "Low Latency",
      subtitle: "MTP 5/1/6"
    }, {
      id: "high-throughput",
      label: "High Throughput",
      subtitle: "Spec decode off"
    }]
  }, {
    id: "quant",
    title: "Quantization",
    options: [{
      id: "fp8",
      label: "FP8"
    }, {
      id: "nvfp4",
      label: "NVFP4",
      disabled: s => !["gb300", "gb200", "b200", "b300"].includes(s.hw),
      disableReason: "The NVFP4 W4A4 kernels are Blackwell-only; Hopper cannot serve this checkpoint."
    }]
  }],
  isRecommendedSelection(s) {
    const pairing = ["h100", "h200"].includes(s.hw) ? "bf16-tilelang" : "fp8-trtllm";
    return s.kvDsaPair === pairing && s.mmTransport === "auto" && s.hicache === "off" && s.bcg === "off" && s.dcp === "off";
  },
  overlayDims: [{
    id: "kvDsaPair",
    title: "KV Cache + DSA Backend",
    default: "fp8-trtllm",
    options: [{
      id: "fp8-trtllm",
      label: "FP8 + TRT-LLM",
      disabled: s => ["h100", "h200"].includes(s.hw),
      disableReason: "FP8 KV cache with TRT-LLM DSA is not supported on Hopper GPUs.",
      stripPrefixes: ["--kv-cache-dtype", "--dsa-prefill-backend", "--dsa-decode-backend"],
      flags: ["--kv-cache-dtype fp8_e4m3", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm"],
      hints: ["Measured on GB300: faster than BF16 + TileLang with about 1.8x the KV token capacity."]
    }, {
      id: "bf16-tilelang",
      label: "BF16 + TileLang",
      stripPrefixes: ["--kv-cache-dtype", "--dsa-prefill-backend", "--dsa-decode-backend"],
      flags: ["--kv-cache-dtype bfloat16", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang"]
    }]
  }, {
    id: "bcg",
    title: "Breakable Cuda Graph",
    default: "off",
    options: [{
      id: "off",
      label: "Off"
    }, {
      id: "on",
      label: "On",
      flags: ["--cuda-graph-backend-prefill breakable"],
      hints: ["Enables breakable prefill CUDA graphs; requires a build with PR #38522."]
    }]
  }, {
    id: "mmTransport",
    title: "VLM Transport",
    default: "auto",
    options: [{
      id: "auto",
      label: "Auto",
      subtitle: "Topology-aware"
    }, {
      id: "cpu",
      label: "CPU",
      subtitle: "Save GPU memory",
      flags: ["--mm-feature-transport cpu"]
    }]
  }, {
    id: "hicache",
    title: "HiCache",
    default: "off",
    options: [{
      id: "off",
      label: "Off"
    }, {
      id: "l2",
      label: "L1 + L2",
      subtitle: "Host memory",
      flags: ["--enable-hierarchical-cache", "--hicache-size 32"],
      hints: ["32 GB host tier; the default ratio can demand more host RAM than the node has free."]
    }, {
      id: "l3",
      label: "+ L3",
      subtitle: "Mooncake",
      flags: ["--enable-hierarchical-cache", "--hicache-size 32", "--hicache-storage-backend mooncake"],
      env: ["SGLANG_HICACHE_MOONCAKE_CONFIG_PATH={{MOONCAKE_CONFIG}}"],
      hints: ["Start Mooncake and place the configuration file on every serving node."]
    }]
  }, {
    id: "dcp",
    title: "Context Parallelism",
    default: "off",
    options: [{
      id: "off",
      label: "Off"
    }, {
      id: "4",
      label: "DCP 4",
      disabled: s => s.hw !== "gb300",
      disableReason: "DCP is validated only on 4x GB300 for now.",
      flags: ["--dcp-size 4", "--dcp-comm-backend a2a", "--dcp-replicate-q-proj"],
      hints: ["Measured on 4x GB300 with both KV/DSA pairings and full decode graph on the earlier TP4/EP4 recipe; re-measurement on the current command is pending."]
    }]
  }],
  modelNames: {
    default: "zai-org/GLM-5.3-Flash",
    nvfp4: "RadixArk/GLM-5.3-Flash-NVFP4"
  },
  placeholders: {
    HOST_IP: {
      target: "command",
      label: "Bind host",
      default: "0.0.0.0"
    },
    PORT: {
      target: "command",
      label: "Bind port",
      default: "30000"
    },
    HF_TOKEN: {
      target: "command",
      label: "HF token (Docker)",
      default: "<your-hf-token>"
    },
    MOONCAKE_CONFIG: {
      target: "command",
      label: "Mooncake config",
      default: "<mooncake.json>"
    },
    CURL_HOST: {
      target: "curl",
      label: "Server host",
      default: "localhost"
    },
    CURL_PORT: {
      target: "curl",
      label: "Server port",
      default: "30000"
    }
  },
  curl: `curl http://{{CURL_HOST}}:{{CURL_PORT}}/v1/chat/completions \\
-H 'Content-Type: application/json' \\
-d '{ "model": "{{MODEL_NAME}}", "messages": [{"role":"user","content":"Hello"}] }'`,
  benchmarkCommands: {
    speed: `# Low Latency speed rows whose notes mention SGLANG_SIMULATE_ACC_LEN=3 were served
# with it to pin the accept length; that number is throughput evidence only. Never
# run accuracy against it.
python3 -m sglang.bench_serving \\
  --backend sglang \\
  --host {{CURL_HOST}} --port {{CURL_PORT}} \\
  --model {{MODEL_NAME}} \\
  --dataset-name {{DATASET}} \\
  --random-input-len {{ISL}} --random-output-len {{OSL}} --random-range-ratio 1.0 \\
  --num-prompts {{NUM_PROMPTS}} --max-concurrency {{MAX_CONCURRENCY}} \\
  --request-rate inf --temperature 0 --seed 42 \\
  --flush-cache`,
    numPromptsByConc: {
      1: 16,
      16: 80,
      64: 320,
      256: 1280,
      1024: 5120
    },
    accuracy: {
      gsm8k_pct: `# To install sgl-eval: pip install sgl-eval
sgl-eval run gsm8k \\
  --base-url http://{{CURL_HOST}}:{{CURL_PORT}}/v1 \\
  --model {{MODEL_NAME}} \\
  --num-threads 64 \\
  --max-tokens 32768 \\
  --temperature 1.0 \\
  --top-p 0.95 \\
  --thinking`
    }
  },
  accuracyLabels: [["gsm8k_pct", "GSM8K", "%"], ["aime2026_pct", "AIME 2026", "%"]],
  dockerImages: {
    gb300: "lmsysorg/sglang:glm-5.3-flash",
    h100: "lmsysorg/sglang:glm-5.3-flash",
    h200: "lmsysorg/sglang:glm-5.3-flash",
    b200: "lmsysorg/sglang:glm-5.3-flash",
    b300: "lmsysorg/sglang:glm-5.3-flash",
    gb200: "lmsysorg/sglang:glm-5.3-flash"
  },
  github: {
    cookbookModel: "zai-org/glm-5.3-flash"
  },
  playgroundFeatures: {
    attention: {
      knobs: [{
        id: "tp",
        label: "TP",
        values: [null, 1, 2, 4, {
          value: 8,
          disable: [{
            when: {
              hw: ["gb300", "gb200"]
            },
            reason: "TP=8 needs 8 GPUs; the GB300 and GB200 recipes run on 4."
          }]
        }]
      }, {
        id: "cp",
        label: "CP",
        values: [null, 1, 2, 4]
      }, {
        id: "dpAttn",
        label: "DP-Attention",
        values: [null, false, 1, 2, 4, {
          value: 8,
          disable: [{
            when: {
              hw: ["gb300", "gb200"]
            },
            reason: "DP-Attention=8 needs 8 ranks; the GB300 and GB200 recipes run on 4."
          }]
        }],
        labels: {
          auto: "Auto",
          false: "Off"
        },
        disable: [{
          when: {
            strategy: ["low-latency"]
          },
          reason: "MTP speculative decoding with DP-Attention is not validated on this model."
        }],
        disableReason: "MTP speculative decoding with DP-Attention is not validated on this model."
      }]
    },
    moe: {
      backend: {
        options: [{
          id: null,
          label: "Inherited"
        }, {
          id: "deep_gemm",
          label: "DeepGemm",
          flags: ["--moe-runner-backend deep_gemm"]
        }]
      },
      ep: {
        label: "EP",
        values: [null, 2, 4, {
          value: 8,
          disable: [{
            when: {
              hw: ["gb300", "gb200"]
            },
            reason: "EP=8 needs 8 GPUs; the GB300 and GB200 recipes run on 4."
          }]
        }]
      }
    },
    parsers: {
      items: [{
        id: "reasoning",
        label: "Reasoning Parser",
        flag: "--reasoning-parser glm45"
      }, {
        id: "toolCall",
        label: "Tool Call Parser",
        flag: "--tool-call-parser glm47"
      }]
    },
    speculative: {
      options: [{
        id: "current",
        label: "Inherited from base"
      }, {
        id: "off",
        label: "Off (greedy)"
      }, {
        id: "eagle",
        label: "EAGLE / MTP 5-1-6",
        flags: ["--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6"],
        disable: [{
          when: {
            dpAttnOn: [true]
          },
          reason: "MTP speculative decoding with DP-Attention is not validated on this model. Turn DP-Attention off in the Attention card above."
        }]
      }, {
        id: "dflash",
        label: "DFlash2",
        flags: ["--speculative-algorithm DFLASH", "--speculative-draft-model-path incoai/GLM-5.3-Flash-DFlash2", "--speculative-draft-attention-backend fa4"],
        note: "⚠️ Needs the GLM-5.3-Flash hidden-state capture from PR #36708. It is merged into the PR #36507 support branch (xinyuan/glm-5.3-flash-support), not into main, so pull that branch at its current head — or add #36708's commit on top of an older checkout — before serving. The lmsysorg/sglang:glm-5.3-flash image alone is not enough.",
        disable: [{
          when: {
            dpAttnOn: [true]
          },
          reason: "DFLASH speculative decoding does not support DP-Attention — the server rejects the combination at startup. Turn DP-Attention off in the Attention card above."
        }]
      }]
    }
  },
  cells: [{
    match: {
      hw: "gb300",
      strategy: "low-latency",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" || s.mmTransport !== "auto" || s.hicache !== "off" ? "unverified" : s.kvDsaPair === "fp8-trtllm" && s.dcp === "off" ? "verified" : ["bf16-tilelang", "fp8-trtllm"].includes(s.kvDsaPair) && ["off", "4"].includes(s.dcp) ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 4", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm", "--kv-cache-dtype fp8_e4m3", "--moe-runner-backend flashinfer_trtllm", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "gb300",
      strategy: "high-throughput",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : ["bf16-tilelang", "fp8-trtllm"].includes(s.kvDsaPair) && s.mmTransport === "auto" && s.hicache === "off" && s.dcp === "off" ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 4", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm", "--kv-cache-dtype fp8_e4m3", "--moe-runner-backend flashinfer_trtllm", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "gb300",
      strategy: "low-latency",
      quant: "nvfp4"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : ["bf16-tilelang", "fp8-trtllm"].includes(s.kvDsaPair) && s.mmTransport === "auto" && s.hicache === "off" && s.dcp === "off" ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--quantization modelopt_fp4", "--tp-size 4", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend flashinfer_cutlass", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--mem-fraction-static 0.85", "--cuda-graph-max-bs-decode 32", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "gb300",
      strategy: "high-throughput",
      quant: "nvfp4"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : ["bf16-tilelang", "fp8-trtllm"].includes(s.kvDsaPair) && s.mmTransport === "auto" && s.hicache === "off" && s.dcp === "off" ? "verified" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--quantization modelopt_fp4", "--tp-size 4", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend flashinfer_cutlass", "--reasoning-parser glm45", "--tool-call-parser glm47", "--mem-fraction-static 0.85", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "gb200",
      strategy: "low-latency",
      quant: "nvfp4"
    },
    nnodes: 1,
    verified: false,
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--quantization modelopt_fp4", "--tp-size 4", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend flashinfer_cutlass", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--mem-fraction-static 0.85", "--cuda-graph-max-bs-decode 32", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "gb200",
      strategy: "high-throughput",
      quant: "nvfp4"
    },
    nnodes: 1,
    verified: false,
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--quantization modelopt_fp4", "--tp-size 4", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend flashinfer_cutlass", "--reasoning-parser glm45", "--tool-call-parser glm47", "--mem-fraction-static 0.85", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "b200",
      strategy: "low-latency",
      quant: "nvfp4"
    },
    nnodes: 1,
    verified: false,
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--quantization modelopt_fp4", "--tp-size 8", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend flashinfer_cutlass", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--mem-fraction-static 0.85", "--cuda-graph-max-bs-decode 32", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "b200",
      strategy: "high-throughput",
      quant: "nvfp4"
    },
    nnodes: 1,
    verified: false,
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--quantization modelopt_fp4", "--tp-size 8", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend flashinfer_cutlass", "--reasoning-parser glm45", "--tool-call-parser glm47", "--mem-fraction-static 0.85", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "b300",
      strategy: "low-latency",
      quant: "nvfp4"
    },
    nnodes: 1,
    verified: false,
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--quantization modelopt_fp4", "--tp-size 8", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend flashinfer_cutlass", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--mem-fraction-static 0.85", "--cuda-graph-max-bs-decode 32", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "b300",
      strategy: "high-throughput",
      quant: "nvfp4"
    },
    nnodes: 1,
    verified: false,
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--quantization modelopt_fp4", "--tp-size 8", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend flashinfer_cutlass", "--reasoning-parser glm45", "--tool-call-parser glm47", "--mem-fraction-static 0.85", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "h100",
      strategy: "low-latency",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : s.mmTransport === "auto" && s.hicache === "off" ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 8", "--ep-size 8", "--mem-fraction-static 0.70", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend deep_gemm", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "h100",
      strategy: "high-throughput",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : ["off", "l2"].includes(s.hicache) ? "verified" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 8", "--ep-size 8", "--mem-fraction-static 0.75", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend deep_gemm", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "h200",
      strategy: "low-latency",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : s.mmTransport === "auto" && s.hicache === "off" ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 8", "--ep-size 8", "--mem-fraction-static 0.75", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend deep_gemm", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "h200",
      strategy: "high-throughput",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : ["off", "l2"].includes(s.hicache) ? "verified" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 8", "--ep-size 8", "--dsa-prefill-backend tilelang", "--dsa-decode-backend tilelang", "--kv-cache-dtype bfloat16", "--moe-runner-backend deep_gemm", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "b200",
      strategy: "low-latency",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : s.hicache === "off" ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 8", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm", "--kv-cache-dtype fp8_e4m3", "--moe-runner-backend flashinfer_trtllm", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "b200",
      strategy: "high-throughput",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : ["off", "l2"].includes(s.hicache) ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 8", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm", "--kv-cache-dtype fp8_e4m3", "--moe-runner-backend flashinfer_trtllm", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "b300",
      strategy: "low-latency",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : s.hicache === "off" ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 8", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm", "--kv-cache-dtype fp8_e4m3", "--moe-runner-backend flashinfer_trtllm", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "b300",
      strategy: "high-throughput",
      quant: "fp8"
    },
    nnodes: 1,
    verified: true,
    verificationStatus: s => s.bcg !== "off" ? "unverified" : ["off", "l2"].includes(s.hicache) ? "in-progress" : "unverified",
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 8", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm", "--kv-cache-dtype fp8_e4m3", "--moe-runner-backend flashinfer_trtllm", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "gb200",
      strategy: "low-latency",
      quant: "fp8"
    },
    nnodes: 1,
    verified: false,
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 4", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm", "--kv-cache-dtype fp8_e4m3", "--moe-runner-backend flashinfer_trtllm", "--speculative-algorithm EAGLE", "--speculative-num-steps 5", "--speculative-eagle-topk 1", "--speculative-num-draft-tokens 6", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }, {
    match: {
      hw: "gb200",
      strategy: "high-throughput",
      quant: "fp8"
    },
    nnodes: 1,
    verified: false,
    env: [],
    flags: ["--model-path {{MODEL_NAME}}", "--tp-size 4", "--dsa-prefill-backend trtllm", "--dsa-decode-backend trtllm", "--kv-cache-dtype fp8_e4m3", "--moe-runner-backend flashinfer_trtllm", "--reasoning-parser glm45", "--tool-call-parser glm47", "--host {{HOST_IP}}", "--port {{PORT}}"]
  }]
};

export const Deployment = ({config, benchmarks}) => {
  if (!config) {
    return <div style={{
      padding: 12,
      color: "#b91c1c"
    }}>Deployment: missing <code>config</code> prop</div>;
  }
  const AMD_RDMA_DOCKER_FLAGS = ["--device /dev/infiniband", "--cap-add IPC_LOCK", "--ulimit memlock=-1", "--ulimit stack=67108864", "--ulimit nofile=1048576:1048576"];
  const HARDWARE_CATALOG = {
    blackwell: [{
      id: "b300",
      label: "B300",
      vram: "288GB"
    }, {
      id: "gb300",
      label: "GB300",
      vram: "288GB"
    }, {
      id: "b200",
      label: "B200",
      vram: "192GB"
    }, {
      id: "gb200",
      label: "GB200",
      vram: "192GB"
    }, {
      id: "dgx-spark",
      label: "DGX Spark",
      vram: "128GB",
      multiNodeDockerFlags: ["--ulimit memlock=-1:-1", "--cap-add IPC_LOCK", "--device /dev/infiniband"]
    }],
    hopper: [{
      id: "h200",
      label: "H200",
      vram: "141GB"
    }, {
      id: "h100",
      label: "H100",
      vram: "80GB"
    }, {
      id: "h20-3e",
      label: "H20-3e",
      vram: "141GB"
    }, {
      id: "h800",
      label: "H800",
      vram: "80GB"
    }],
    amd: [{
      id: "mi300x",
      label: "MI300X",
      vram: "192GB",
      multiNodeDockerFlags: [...AMD_RDMA_DOCKER_FLAGS]
    }, {
      id: "mi325x",
      label: "MI325X",
      vram: "256GB",
      multiNodeDockerFlags: [...AMD_RDMA_DOCKER_FLAGS]
    }, {
      id: "mi350x",
      label: "MI350X",
      vram: "288GB",
      multiNodeDockerFlags: [...AMD_RDMA_DOCKER_FLAGS]
    }, {
      id: "mi355x",
      label: "MI355X",
      vram: "288GB",
      multiNodeDockerFlags: [...AMD_RDMA_DOCKER_FLAGS]
    }],
    npu: [{
      id: "a3",
      label: "Ascend A3 Series",
      vram: "64GB/die"
    }]
  };
  const makeStyles = isDark => ({
    container: {
      maxWidth: "900px",
      margin: "0 auto",
      display: "flex",
      flexDirection: "column",
      gap: "3px"
    },
    card: {
      padding: "5px 10px",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      borderLeft: `3px solid ${isDark ? "#E85D4D" : "#D45D44"}`,
      borderRadius: "4px",
      display: "flex",
      alignItems: "center",
      gap: "10px",
      background: isDark ? "#1f2937" : "#fff"
    },
    cardColumn: {
      padding: "5px 10px",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      borderLeft: `3px solid ${isDark ? "#E85D4D" : "#D45D44"}`,
      borderRadius: "4px",
      display: "flex",
      flexDirection: "column",
      gap: "4px",
      background: isDark ? "#1f2937" : "#fff"
    },
    title: {
      fontSize: "12px",
      fontWeight: "600",
      minWidth: "108px",
      flexShrink: 0,
      color: isDark ? "#e5e7eb" : "inherit"
    },
    vendorRow: {
      display: "flex",
      alignItems: "center",
      gap: "6px"
    },
    vendorLabel: {
      fontSize: "10px",
      fontWeight: "600",
      color: isDark ? "#9ca3af" : "#6b7280",
      width: "68px",
      flexShrink: 0,
      textTransform: "uppercase",
      letterSpacing: "0.04em"
    },
    itemsGrid: () => ({
      display: "grid",
      gridTemplateColumns: "repeat(auto-fit, minmax(72px, 1fr))",
      gap: "4px",
      flex: 1
    }),
    labelBase: {
      padding: "2px 8px",
      border: `1px solid ${isDark ? "#9ca3af" : "#d1d5db"}`,
      borderRadius: "3px",
      cursor: "pointer",
      display: "inline-flex",
      flexDirection: "column",
      alignItems: "center",
      justifyContent: "center",
      fontWeight: "500",
      fontSize: "12px",
      transition: "all 0.2s",
      userSelect: "none",
      minHeight: "26px",
      textAlign: "center",
      background: isDark ? "#374151" : "#fff",
      color: isDark ? "#e5e7eb" : "inherit"
    },
    checked: {
      background: "#D45D44",
      color: "white",
      borderColor: "#D45D44"
    },
    disabled: {
      cursor: "not-allowed",
      opacity: 0.4
    },
    subtitle: {
      display: "block",
      fontSize: "9px",
      marginTop: "1px",
      lineHeight: "1.1",
      opacity: 0.7
    },
    commandWrap: {
      position: "relative",
      flex: 1,
      background: isDark ? "#111827" : "#f5f5f5",
      borderRadius: "6px",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      overflow: "hidden"
    },
    commandHeader: {
      display: "flex",
      flexWrap: "wrap",
      justifyContent: "space-between",
      alignItems: "center",
      gap: "6px 10px",
      padding: "6px 10px",
      borderBottom: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      background: isDark ? "#1f2937" : "#fafafa"
    },
    commandPre: {
      padding: "12px 16px",
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace",
      fontSize: "12px",
      lineHeight: "1.5",
      color: isDark ? "#e5e7eb" : "#374151",
      whiteSpace: "pre-wrap",
      overflowX: "auto",
      margin: 0
    },
    mtpWarn: {
      margin: "8px 0 0",
      padding: "8px 12px",
      borderRadius: "8px",
      fontSize: "12px",
      lineHeight: "1.45",
      background: isDark ? "#78350f" : "#fef3c7",
      color: isDark ? "#fde68a" : "#92400e",
      border: `1px solid ${isDark ? "#92400e" : "#fcd34d"}`
    },
    badge: status => ({
      display: "inline-flex",
      alignItems: "center",
      gap: "6px",
      padding: "2px 8px",
      borderRadius: "10px",
      background: ({
        verified: isDark ? "#064e3b" : "#d1fae5",
        "in-progress": isDark ? "#1e3a8a" : "#dbeafe",
        unverified: isDark ? "#78350f" : "#fef3c7"
      })[verifyStatusOf(status)],
      color: ({
        verified: isDark ? "#a7f3d0" : "#065f46",
        "in-progress": isDark ? "#bfdbfe" : "#1e40af",
        unverified: isDark ? "#fde68a" : "#92400e"
      })[verifyStatusOf(status)],
      fontSize: "11px",
      fontWeight: 600,
      whiteSpace: "nowrap"
    }),
    badgeDot: status => ({
      width: "8px",
      height: "8px",
      borderRadius: "50%",
      background: ({
        verified: "#10b981",
        "in-progress": "#3b82f6",
        unverified: "#f59e0b"
      })[verifyStatusOf(status)]
    }),
    iconButton: {
      padding: "4px 10px",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "4px",
      background: isDark ? "#1f2937" : "#fff",
      color: isDark ? "#e5e7eb" : "#374151",
      fontSize: "11px",
      fontWeight: 500,
      cursor: "pointer",
      display: "inline-flex",
      alignItems: "center",
      gap: "4px"
    },
    iconRow: {
      display: "inline-flex",
      flexWrap: "wrap",
      gap: "6px"
    },
    runModeWrap: {
      display: "inline-flex",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "10px",
      overflow: "hidden",
      fontSize: "11px",
      fontWeight: 600,
      userSelect: "none"
    },
    runModeChip: active => ({
      padding: "2px 10px",
      cursor: "pointer",
      background: active ? isDark ? "#1f2937" : "#fff" : "transparent",
      color: active ? isDark ? "#e5e7eb" : "#111827" : isDark ? "#9ca3af" : "#6b7280",
      borderRight: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`
    }),
    runModeChipLast: active => ({
      padding: "2px 10px",
      cursor: "pointer",
      background: active ? isDark ? "#1f2937" : "#fff" : "transparent",
      color: active ? isDark ? "#e5e7eb" : "#111827" : isDark ? "#9ca3af" : "#6b7280"
    }),
    headerLeft: {
      display: "inline-flex",
      flexWrap: "wrap",
      alignItems: "center",
      gap: "8px"
    },
    modalBackdrop: {
      position: "fixed",
      inset: 0,
      background: "rgba(0,0,0,0.5)",
      display: "flex",
      alignItems: "center",
      justifyContent: "center",
      zIndex: 9999
    },
    modalBox: {
      background: isDark ? "#1f2937" : "#fff",
      color: isDark ? "#e5e7eb" : "#111827",
      borderRadius: "8px",
      padding: "20px",
      maxWidth: "720px",
      width: "92%",
      maxHeight: "85vh",
      overflowY: "auto",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      boxShadow: "0 10px 25px rgba(0,0,0,0.25)"
    },
    modalHeader: {
      display: "flex",
      justifyContent: "space-between",
      alignItems: "center",
      marginBottom: "12px"
    },
    modalTitle: {
      fontSize: "15px",
      fontWeight: 600
    },
    modalCloseBtn: {
      background: "transparent",
      border: "none",
      color: "inherit",
      fontSize: "20px",
      cursor: "pointer",
      padding: "0 6px",
      lineHeight: 1
    },
    formField: {
      display: "flex",
      flexDirection: "column",
      gap: "4px",
      marginBottom: "10px"
    },
    formLabel: {
      fontSize: "12px",
      fontWeight: 500,
      color: isDark ? "#9ca3af" : "#4b5563"
    },
    formInput: {
      padding: "6px 10px",
      fontSize: "13px",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "4px",
      background: isDark ? "#111827" : "#fff",
      color: isDark ? "#e5e7eb" : "#111827",
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace"
    },
    sectionHeading: {
      fontSize: "12px",
      fontWeight: 600,
      textTransform: "uppercase",
      letterSpacing: "0.04em",
      color: isDark ? "#9ca3af" : "#6b7280",
      margin: "12px 0 6px 0"
    },
    primaryBtn: {
      padding: "6px 14px",
      background: "#D45D44",
      color: "white",
      border: "none",
      borderRadius: "4px",
      cursor: "pointer",
      fontSize: "13px",
      fontWeight: 500
    },
    benchCard: {
      padding: "8px 12px",
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      borderLeft: `3px solid ${isDark ? "#E85D4D" : "#D45D44"}`,
      borderRadius: "4px",
      background: isDark ? "#1f2937" : "#fff",
      display: "flex",
      flexDirection: "column",
      gap: "8px"
    },
    benchHeader: {
      display: "flex",
      flexWrap: "wrap",
      alignItems: "baseline",
      justifyContent: "space-between",
      gap: "6px 12px"
    },
    benchTitle: {
      fontSize: "13px",
      fontWeight: 600,
      color: isDark ? "#e5e7eb" : "inherit"
    },
    benchVersion: {
      fontSize: "11px",
      color: isDark ? "#9ca3af" : "#6b7280"
    },
    benchHeaderRight: {
      display: "flex",
      flexWrap: "wrap",
      alignItems: "center",
      gap: "6px 10px",
      flexShrink: 0
    },
    benchChipRow: {
      display: "flex",
      alignItems: "center",
      gap: "6px",
      flexWrap: "wrap",
      margin: "2px 0 8px"
    },
    benchChip: {
      padding: "2px 10px",
      fontSize: "12px",
      cursor: "pointer",
      border: `1px solid ${isDark ? "#4b5563" : "#d1d5db"}`,
      borderRadius: "4px",
      background: isDark ? "#1f2937" : "#fff",
      color: isDark ? "#e5e7eb" : "#374151",
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace"
    },
    benchChipActive: {
      background: "#D45D44",
      color: "white",
      borderColor: "#D45D44"
    },
    benchBlock: {
      border: `1px solid ${isDark ? "#374151" : "#e5e7eb"}`,
      borderRadius: "4px",
      padding: "8px 10px",
      background: isDark ? "#111827" : "#fafafa"
    },
    benchBlockTitle: {
      fontSize: "11px",
      fontWeight: 600,
      textTransform: "uppercase",
      letterSpacing: "0.04em",
      color: isDark ? "#9ca3af" : "#6b7280",
      marginBottom: "4px"
    },
    benchWorkload: {
      fontSize: "11px",
      fontStyle: "italic",
      color: isDark ? "#9ca3af" : "#6b7280",
      marginBottom: "6px",
      lineHeight: "1.3"
    },
    benchRow: {
      display: "flex",
      justifyContent: "space-between",
      fontSize: "12px",
      padding: "2px 0"
    },
    benchKey: {
      color: isDark ? "#9ca3af" : "#6b7280"
    },
    benchVal: {
      color: isDark ? "#e5e7eb" : "#111827",
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace",
      fontWeight: 500
    },
    benchNotes: {
      fontSize: "11px",
      fontStyle: "italic",
      color: isDark ? "#9ca3af" : "#6b7280"
    },
    benchLegend: {
      fontSize: "10px",
      fontStyle: "italic",
      color: isDark ? "#6b7280" : "#9ca3af",
      marginTop: "6px",
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace"
    },
    benchEmpty: {
      fontSize: "12px",
      fontStyle: "italic",
      color: isDark ? "#9ca3af" : "#6b7280"
    },
    benchTable: {
      display: "grid",
      columnGap: 0,
      rowGap: "3px",
      marginTop: "4px",
      alignItems: "baseline"
    },
    benchTableHead: {
      textAlign: "right",
      fontWeight: 500,
      fontSize: "11px",
      color: isDark ? "#9ca3af" : "#6b7280",
      paddingLeft: "16px",
      paddingBottom: "4px",
      whiteSpace: "nowrap"
    },
    benchTableCornerHead: {
      paddingBottom: "4px"
    },
    benchTableSeparator: {
      gridColumn: "1 / -1",
      height: "1px",
      background: isDark ? "#374151" : "#e5e7eb",
      marginTop: "-3px"
    },
    benchTableLabel: {
      textAlign: "left",
      fontSize: "12px",
      color: isDark ? "#9ca3af" : "#6b7280",
      whiteSpace: "nowrap"
    },
    benchTableValue: {
      textAlign: "right",
      fontSize: "12px",
      color: isDark ? "#e5e7eb" : "#111827",
      fontFamily: "'Menlo', 'Monaco', 'Courier New', monospace",
      fontWeight: 500,
      paddingLeft: "16px",
      whiteSpace: "nowrap"
    },
    benchTableValueMissing: {
      color: isDark ? "#6b7280" : "#9ca3af"
    }
  });
  const VERIFY_LABEL = {
    verified: "Verified",
    "in-progress": "Final Verification In Progress",
    unverified: "Not Verified"
  };
  const verifyStatusOf = v => typeof v === "string" ? VERIFY_LABEL[v] ? v : "unverified" : v ? "verified" : "unverified";
  const cellVerifyStatus = (c, sel) => {
    if (!c) return "unverified";
    const v = typeof c.verificationStatus === "function" ? c.verificationStatus(sel) : c.verificationStatus;
    return verifyStatusOf(v ?? c.verified);
  };
  const LEGACY_MATCH_DIMS = [{
    id: "variant",
    title: "Model Variant",
    optionsKey: "variants"
  }, {
    id: "quant",
    title: "Quantization",
    optionsKey: "quantizations"
  }, {
    id: "strategy",
    title: "Strategy",
    optionsKey: "strategies"
  }, {
    id: "nodes",
    title: "Nodes",
    optionsKey: "nodesOptions"
  }];
  const matchDimSpecs = (config.matchDims || LEGACY_MATCH_DIMS).map(d => ({
    ...d,
    options: d.options || config[d.optionsKey] || []
  }));
  const overlayDimSpecs = config.overlayDims || [];
  const commandBuilder = config.commandBuilder || null;
  const DIMENSIONS = ["hw", ...matchDimSpecs.map(d => d.id)];
  const optionVisible = (opt, sel) => typeof opt.showWhen !== "function" || opt.showWhen(sel);
  const optionDisabled = (opt, sel) => typeof opt.disabled === "function" ? opt.disabled(sel) : !!opt.disabled;
  const visibleOptions = (spec, sel) => (spec.options || []).filter(o => optionVisible(o, sel));
  const rowVisible = (spec, sel) => (typeof spec.showWhen !== "function" || spec.showWhen(sel)) && visibleOptions(spec, sel).length > 0;
  const overlayPick = sel => {
    const picked = [];
    for (const spec of config.overlayDims || []) {
      if (!rowVisible(spec, sel)) continue;
      const opt = (spec.options || []).find(o => o.id === sel[spec.id]);
      if (opt && !optionDisabled(opt, sel)) picked.push(opt);
    }
    return picked;
  };
  const overlayPart = (sel, key) => {
    const out = [];
    for (const opt of overlayPick(sel)) {
      const add = typeof opt[key] === "function" ? opt[key](sel) : opt[key];
      if (add) out.push(...add);
    }
    return out;
  };
  const overlayCompose = (cellFlags, sel) => {
    const strip = overlayPart(sel, "stripPrefixes");
    const add = overlayPart(sel, "flags");
    if (!strip.length) return [...cellFlags || [], ...add];
    const used = new Set();
    const replacementsFor = tok => {
      const out = [];
      add.forEach((f, i) => {
        if (used.has(i) || f.split(/[\s=]/)[0] !== tok) return;
        used.add(i);
        out.push(f);
      });
      return out;
    };
    const out = [];
    for (const f of cellFlags || []) {
      const tok = f.split(/[\s=]/)[0];
      if (!strip.includes(tok)) out.push(f); else out.push(...replacementsFor(tok));
    }
    add.forEach((f, i) => {
      if (!used.has(i)) out.push(f);
    });
    return out;
  };
  const optionSoft = (opt, sel) => typeof opt.soft === "function" ? opt.soft(sel) : !!opt.soft;
  const findCell = (cells, sel) => cells.find(c => DIMENSIONS.every(d => c.match[d] === sel[d]));
  const findBenchmark = (list, sel) => {
    const hits = (list || []).filter(b => Object.entries(b.match || ({})).every(([k, v]) => sel[k] === v));
    return hits.sort((a, b) => Object.keys(b.match).length - Object.keys(a.match).length)[0] || null;
  };
  const normalizeSpeed = speed => {
    if (!speed) return [];
    return Array.isArray(speed) ? speed : [speed];
  };
  const effectiveAccuracy = (entry, sel) => entry ? {
    ...config.defaultAccuracy && config.defaultAccuracy[sel.variant] || ({}),
    ...entry.accuracy || ({})
  } : {};
  const benchmarkIsEmpty = (entry, accuracy) => {
    for (const m of normalizeSpeed(entry && entry.speed)) {
      if (m && typeof m === "object") {
        for (const [key, v] of Object.entries(m)) {
          if (key === "workload") continue;
          if (v !== null && v !== undefined) return false;
        }
      }
    }
    if (accuracy && typeof accuracy === "object") {
      for (const v of Object.values(accuracy)) {
        if (v !== null && v !== undefined) return false;
      }
    }
    return true;
  };
  const isOptionAvailable = (cells, sel, dim, value) => {
    const idx = DIMENSIONS.indexOf(dim);
    const higher = DIMENSIONS.slice(0, idx);
    return cells.some(c => c.match[dim] === value && higher.every(d => c.match[d] === sel[d]));
  };
  const snapToValidCell = (cells, sel, dim, value) => {
    const idx = DIMENSIONS.indexOf(dim);
    const higher = DIMENSIONS.slice(0, idx);
    const lower = DIMENSIONS.slice(idx + 1);
    let best = null, bestLowerMatches = -1;
    for (const c of cells) {
      if (c.match[dim] !== value) continue;
      if (!higher.every(d => c.match[d] === sel[d])) continue;
      let s = 0;
      for (const d of lower) if (c.match[d] === sel[d]) s++;
      if (s > bestLowerMatches) {
        bestLowerMatches = s;
        best = c;
      }
    }
    if (!best) return sel;
    const next = {
      ...sel,
      [dim]: value
    };
    for (const d of lower) next[d] = best.match[d];
    return next;
  };
  const validateSelection = (cells, parsed) => {
    const valid = {};
    for (const dim of DIMENSIONS) {
      const want = parsed[dim];
      const works = cells.some(c => c.match[dim] === want && DIMENSIONS.slice(0, DIMENSIONS.indexOf(dim)).every(d => c.match[d] === valid[d]));
      if (works) {
        valid[dim] = want;
      } else {
        const fallback = cells.find(c => DIMENSIONS.slice(0, DIMENSIONS.indexOf(dim)).every(d => c.match[d] === valid[d]));
        valid[dim] = fallback ? fallback.match[dim] : want;
      }
    }
    for (const spec of overlayDimSpecs) {
      const want = parsed[spec.id];
      const opts = spec.options || [];
      const picked = opts.some(o => o.id === want) ? want : (spec.default ?? (opts[0] && opts[0].id)) ?? "";
      const withPick = {
        ...valid,
        [spec.id]: picked
      };
      const usable = visibleOptions(spec, withPick).filter(o => !optionDisabled(o, withPick));
      valid[spec.id] = usable.some(o => o.id === picked) ? picked : (usable[0] && usable[0].id) ?? picked;
    }
    return valid;
  };
  const resolveModelName = sel => {
    const keys = [`${sel.hw}|${sel.variant}|${sel.quant}`, `${sel.variant}|${sel.quant}`, `${sel.hw}|${sel.quant}`, sel.quant, sel.hw, "default"];
    for (const k of keys) {
      const hit = config.modelNames[k];
      if (hit) return hit;
    }
    return "";
  };
  const interpolate = (text, env, modelName) => text.replace(/{{(\w+)}}/g, (_, key) => key === "MODEL_NAME" ? modelName : env[key] ?? `{{${key}}}`);
  const parseNnodes = id => {
    if (Number.isInteger(id)) return id;
    if ((/^\d+$/).test(id || "")) return parseInt(id, 10);
    if (id === "single") return 1;
    const m = (/^multi-(\d+)$/).exec(id || "");
    return m ? parseInt(m[1], 10) : 1;
  };
  const cellNnodes = (cell, sel) => sel.nodes !== undefined ? parseNnodes(sel.nodes) : cell.nnodes || 1;
  const PD_SERVE_PORTS = {
    prefill: 30000,
    decode: 30100
  };
  const overlayEnv = sel => overlayPart(sel, "env");
  const overlayHints = sel => overlayPart(sel, "hints");
  const renderCommand = (cell, sel, envValues, mode = "python") => {
    if (!cell) return "# No command available for the current selection.";
    const modelName = resolveModelName(sel);
    const nnodes = cellNnodes(cell, sel);
    const multinode = nnodes > 1;
    const cellEnv = [...cell.env || [], ...overlayEnv(sel)];
    const flags = overlayCompose(cell.flags, sel);
    if (multinode) {
      const PARALLELISM_ANCHORS = new Set(["--enable-dp-attention", "--dp-size", "--dp", "--tp-size", "--tp", "--sp-degree", "--ulysses-degree", "--ring-degree"]);
      let i = flags.reduce((last, flag, index) => PARALLELISM_ANCHORS.has(flag.split(/[\s=]/)[0]) ? index : last, -1);
      if (i === -1) i = flags.findIndex(f => f.startsWith("--model-path"));
      flags.splice(i + 1, 0, `--nnodes ${nnodes}`, `--node-rank {{NODE_RANK}}`, `--dist-init-addr {{NODE0_IP}}:20000`);
    }
    const pdServePort = PD_SERVE_PORTS[sel.pdMode];
    if (pdServePort !== undefined) {
      for (let j = 0; j < flags.length; j++) {
        if (flags[j].split(/[\s=]/)[0] === "--port") {
          flags[j] = `--port ${pdServePort}`;
        }
      }
    }
    let cmd;
    if (mode === "docker") {
      const di = config.dockerImages || ({});
      const image = di[`${sel.hw}|${sel.variant}|${sel.quant}`] || di[`${sel.variant}|${sel.quant}`] || di[`${sel.hw}|${sel.quant}|${sel.strategy}`] || di[`${sel.hw}|${sel.quant}`] || di[sel.hw] || "lmsysorg/sglang:dev";
      const dockerRunCommand = typeof config.dockerRunCommand === "function" ? config.dockerRunCommand(sel) : config.dockerRunCommand || "sglang serve";
      const portFlag = flags.find(x => x.split(/[\s=]/)[0] === "--port");
      const servePort = portFlag ? portFlag.slice(("--port").length).trim() : "{{PORT}}";
      const hostNetwork = multinode || typeof config.dockerHostNetworkWhen === "function" && config.dockerHostNetworkWhen(sel, {
        flags,
        env: cellEnv
      });
      const vendorOf = hwId => {
        for (const [vendor, list] of Object.entries(HARDWARE_CATALOG)) {
          if (list.some(h => h.id === hwId)) return vendor;
        }
        const extra = (config.hardware || []).find(h => h.id === hwId);
        return extra && extra.vendor || "nvidia";
      };
      const fabricFlagsOf = hwId => {
        const extra = (config.hardware || []).find(h => h.id === hwId);
        if (extra) return extra.multiNodeDockerFlags || [];
        for (const list of Object.values(HARDWARE_CATALOG)) {
          const hit = list.find(h => h.id === hwId);
          if (hit) return hit.multiNodeDockerFlags || [];
        }
        return [];
      };
      const gpuAccessLines = vendorOf(sel.hw) === "amd" ? ["docker run", "  --device=/dev/kfd --device=/dev/dri", "  --group-add video", "  --cap-add=SYS_PTRACE --security-opt seccomp=unconfined", "  --shm-size 32g"] : vendorOf(sel.hw) === "npu" ? ["docker run --privileged --shm-size=16g", "  --device=/dev/davinci0 --device=/dev/davinci1 --device=/dev/davinci2 --device=/dev/davinci3", "  --device=/dev/davinci4 --device=/dev/davinci5 --device=/dev/davinci6 --device=/dev/davinci7", "  --device=/dev/davinci8 --device=/dev/davinci9 --device=/dev/davinci10 --device=/dev/davinci11", "  --device=/dev/davinci12 --device=/dev/davinci13 --device=/dev/davinci14 --device=/dev/davinci15", "  --device=/dev/davinci_manager", "  --device=/dev/hisi_hdc", "  -v /usr/local/sbin:/usr/local/sbin", "  -v /usr/local/Ascend/driver:/usr/local/Ascend/driver", "  -v /usr/local/Ascend/firmware:/usr/local/Ascend/firmware", "  -v /etc/ascend_install.info:/etc/ascend_install.info", "  -v /var/queue_schedule:/var/queue_schedule", "  -v ~/.cache/:/root/.cache/"] : ["docker run --gpus all", "  --shm-size 32g"];
      const dockerLines = [...gpuAccessLines, hostNetwork ? "  --network host" : `  -p ${servePort}:${servePort}`, ...multinode ? fabricFlagsOf(sel.hw).map(f => "  " + f) : [], ...vendorOf(sel.hw) === "npu" ? [] : ["  -v ~/.cache/huggingface:/root/.cache/huggingface"], ...(config.dockerMounts || []).map(mount => `  -v ${mount}`), ...config.placeholders && config.placeholders.HF_TOKEN ? [`  --env "HF_TOKEN={{HF_TOKEN}}"`] : [], ...cellEnv.map(e => `  --env ${e}`), "  --ipc=host", `  ${image}`, `  ${dockerRunCommand}`, ...flags.map(f => "    " + f)];
      cmd = dockerLines.join(" \\\n");
    } else {
      const flagBlock = flags.map(f => "  " + f).join(" \\\n");
      const envBlock = cellEnv.length ? cellEnv.join(" \\\n") + " \\\n" : "";
      cmd = `${envBlock}sglang serve \\\n${flagBlock}`;
    }
    const hintLines = [...overlayHints(sel), ...multinode && config.multiNodeHints && config.multiNodeHints[sel.hw] ? config.multiNodeHints[sel.hw] : []];
    if (hintLines.length) {
      const hint = hintLines.map(line => line.length ? "# " + line : "#").join("\n");
      cmd = `${hint}\n${cmd}`;
    }
    cmd = interpolate(cmd, envValues, modelName);
    if (multinode) {
      const header = `# Multi-node (${nnodes} nodes). Run the same command on every node with:\n` + `#   <node-rank> = 0 on the head node, 1..${nnodes - 1} on the others\n` + `#   <node0-ip>  = IP of the head node (reachable from all others)`;
      cmd = `${header}\n${cmd}`;
    }
    return cmd;
  };
  const ACCURACY_LABELS = config.accuracyLabels || [];
  const renderBenchmarkCard = entry => {
    const pct = entry && entry.latencyPercentile || config.latencyPercentile || "P50";
    const SPEED_LABELS = [["ttft_ms", `TTFT (${pct})`, "ms"], ["tpot_ms", `TPOT (${pct})`, "ms"], ["tokens_per_sec_per_gpu", "throughput per gpu", "tok/s"], ["interactivity", "interactivity", "tokens/s/user", m => m.tpot_ms != null && m.tpot_ms !== 0 ? Math.round(1000 / m.tpot_ms * 10) / 10 : null]];
    const WORKLOAD_KEYS = ["dataset", "isl", "osl", "max_concurrency"];
    const fmt = (val, unit) => {
      if (val === null || val === undefined) return null;
      return `${val}${unit ? " " + unit : ""}`;
    };
    const formatWorkloadParts = (workload, keys) => {
      if (!workload) return "";
      const parts = [];
      if (keys.has("dataset") && workload.dataset) parts.push(workload.dataset);
      if (keys.has("isl") || keys.has("osl")) {
        if (workload.isl != null || workload.osl != null) {
          parts.push(`in/out=${workload.isl != null ? workload.isl : "?"}/${workload.osl != null ? workload.osl : "?"}`);
        }
      }
      if (keys.has("max_concurrency") && workload.max_concurrency != null) {
        parts.push(`max-concurrency=${workload.max_concurrency}`);
      }
      return parts.join(", ");
    };
    const ALWAYS_PER_COLUMN = new Set(["max_concurrency"]);
    const partitionWorkload = measurements => {
      const shared = new Set();
      const differing = new Set();
      for (const k of WORKLOAD_KEYS) {
        const seen = new Set();
        let anyPresent = false;
        for (const m of measurements) {
          const v = m && m.workload ? m.workload[k] : undefined;
          if (v != null) anyPresent = true;
          seen.add(v);
        }
        if (!anyPresent) continue;
        if (ALWAYS_PER_COLUMN.has(k) || seen.size > 1) differing.add(k); else shared.add(k);
      }
      return {
        shared,
        differing
      };
    };
    const renderBenchTable = ({title, sharedText, colHeaders, rows, colCount, legend}) => {
      if (rows.length === 0) return null;
      const showColHeaders = colHeaders.length > 0 && colHeaders.some(h => h !== "");
      return <div style={s.benchBlock}>
          <div style={s.benchBlockTitle}>{title}</div>
          {sharedText && <div style={s.benchWorkload}>{sharedText}</div>}
          <div style={{
        ...s.benchTable,
        gridTemplateColumns: `max-content repeat(${colCount}, minmax(0, 1fr))`
      }}>
            {showColHeaders && <div key="corner" style={s.benchTableCornerHead}></div>}
            {showColHeaders && colHeaders.map((h, i) => <div key={`hdr-${i}`} style={s.benchTableHead}>{h}</div>)}
            {showColHeaders && <div key="sep" style={s.benchTableSeparator}></div>}
            {rows.map(r => [<div key={`lbl-${r.label}`} style={s.benchTableLabel}>{r.label}</div>, ...r.values.map((v, i) => <div key={`val-${r.label}-${i}`} style={v === null ? {
        ...s.benchTableValue,
        ...s.benchTableValueMissing
      } : s.benchTableValue}>
                  {v !== null ? v : "—"}
                </div>)])}
          </div>
          {legend && <div style={s.benchLegend}>
              {(Array.isArray(legend) ? legend : [legend]).map((line, i) => <div key={`legend-${i}`}>{line}</div>)}
            </div>}
        </div>;
    };
    const buildSpeedTable = measurements => {
      if (measurements.length === 0) return null;
      const {shared, differing} = partitionWorkload(measurements);
      const sharedText = formatWorkloadParts(measurements[0] && measurements[0].workload, shared);
      const colHeaders = measurements.map(m => formatWorkloadParts(m && m.workload, differing));
      const rows = SPEED_LABELS.map(tup => {
        const [key, label, unit, compute] = tup;
        const values = measurements.map(m => {
          const raw = compute ? compute(m) : m[key];
          return fmt(raw, unit);
        });
        return {
          label,
          values
        };
      });
      return {
        title: "Speed",
        sharedText,
        colHeaders,
        rows,
        colCount: measurements.length,
        legend: [`throughput per gpu = (input+output tokens)/elapsed/GPU`, `interactivity = 1000/TPOT(ms) (tokens/s/user)`]
      };
    };
    const buildAccuracyTable = accuracy => {
      if (!accuracy) return null;
      const rows = ACCURACY_LABELS.map(([key, label, unit]) => {
        const v = fmt(accuracy[key], unit);
        if (v === null) return null;
        return {
          label,
          values: [v]
        };
      }).filter(r => r !== null);
      if (rows.length === 0) return null;
      return {
        title: "Accuracy",
        sharedText: null,
        colHeaders: [],
        rows,
        colCount: 1
      };
    };
    const accuracy = effectiveAccuracy(entry, sel);
    const isEmpty = benchmarkIsEmpty(entry, accuracy);
    const measurements = !isEmpty ? normalizeSpeed(entry && entry.speed) : [];
    const accuracyTable = !isEmpty ? buildAccuracyTable(accuracy) : null;
    const speedTable = !isEmpty ? buildSpeedTable(measurements) : null;
    const hasBenchCmds = !isEmpty && buildBenchCommands(entry, sel) !== null;
    return <div style={s.benchCard}>
        <div style={s.benchHeader}>
          <div style={s.benchTitle}>Benchmark</div>
          <div style={s.benchHeaderRight}>
            {!isEmpty && entry && entry.sglang_version && <div style={s.benchVersion}>measured on sglang <code>{entry.sglang_version}</code></div>}
            {hasBenchCmds && <button style={s.iconButton} onClick={() => setModal("bench")}>⚡ Reproduce</button>}
          </div>
        </div>
        {isEmpty ? <div style={s.benchEmpty}>
            Benchmark data pending for this combination — submit yours via the Playground's Submit ↗ button.
          </div> : <>
            {accuracyTable && renderBenchTable(accuracyTable)}
            {speedTable && renderBenchTable(speedTable)}
            {entry && entry.notes && <div style={s.benchNotes}>{entry.notes}</div>}
          </>}
      </div>;
  };
  const buildBenchCommands = (entry, sel) => {
    const bc = config.benchmarkCommands;
    if (!bc) return null;
    const acc = effectiveAccuracy(entry, sel);
    const accuracy = [];
    if (bc.accuracy) {
      for (const [key, label] of ACCURACY_LABELS) {
        if (acc[key] == null) continue;
        const tmpl = bc.accuracy[key];
        const resolved = typeof tmpl === "string" ? tmpl : tmpl && tmpl[sel.variant] || null;
        if (resolved) accuracy.push({
          key,
          label,
          template: resolved
        });
      }
    }
    let speed = null;
    if (bc.speed && entry) {
      const ms = normalizeSpeed(entry.speed).filter(m => m && m.workload && m.workload.max_concurrency != null);
      const concurrencies = [...new Set(ms.map(m => m.workload.max_concurrency))].sort((a, b) => a - b);
      if (concurrencies.length) {
        speed = {
          template: bc.speed,
          concurrencies,
          workload: ms[0].workload,
          numPromptsOf: c => {
            const m = ms.find(x => x.workload.max_concurrency === c);
            if (m && m.workload.num_prompts != null) return m.workload.num_prompts;
            const tbl = bc.numPromptsByConc;
            if (tbl && tbl[c] != null) return tbl[c];
            return Math.max(c * 2, 200);
          }
        };
      }
    }
    if (accuracy.length === 0 && !speed) return null;
    return {
      accuracy,
      speed
    };
  };
  const buildHardwareGroups = () => {
    const supported = new Set(config.supportedHardware);
    const catalog = {};
    for (const [vendor, list] of Object.entries(HARDWARE_CATALOG)) catalog[vendor] = [...list];
    for (const hw of config.hardware || []) {
      const vendor = hw.vendor || "nvidia";
      const list = catalog[vendor] || (catalog[vendor] = []);
      const entry = {
        id: hw.id,
        label: hw.label,
        vram: hw.vram
      };
      const i = list.findIndex(x => x.id === hw.id);
      if (i >= 0) list[i] = entry; else list.push(entry);
    }
    const groups = [];
    for (const [vendor, list] of Object.entries(catalog)) {
      const items = list.filter(hw => supported.has(hw.id)).map(hw => ({
        id: hw.id,
        label: hw.label,
        subtitle: hw.vram
      }));
      if (items.length) groups.push({
        label: vendor.toUpperCase(),
        items
      });
    }
    if (config.groupHardware === false) {
      return [{
        label: null,
        items: groups.flatMap(group => group.items)
      }];
    }
    return groups;
  };
  const initialSelectionFromCells = () => {
    const first = (config.cells || [])[0];
    const sel = Object.fromEntries(DIMENSIONS.map(d => [d, first ? first.match[d] : ""]));
    for (const spec of overlayDimSpecs) {
      const opts = spec.options || [];
      sel[spec.id] = (spec.default ?? (opts[0] && opts[0].id)) ?? "";
    }
    if (!commandBuilder) return sel;
    return {
      ...sel,
      hw: commandBuilder.defaultSelection?.hw || config.supportedHardware?.[0] || "",
      ...commandBuilder.defaultSelection || ({})
    };
  };
  const normalizeBuilderSelection = parsed => {
    const out = {
      ...initialSelectionFromCells(),
      ...parsed
    };
    if (!(config.supportedHardware || []).includes(out.hw)) {
      out.hw = commandBuilder.defaultSelection?.hw || config.supportedHardware?.[0] || "";
    }
    for (const spec of overlayDimSpecs) {
      if (spec.kind === "number") {
        const value = Number.parseInt(out[spec.id], 10);
        out[spec.id] = Math.min(spec.max, Math.max(spec.min, Number.isFinite(value) ? value : Number(spec.default ?? spec.min)));
        continue;
      }
      const options = spec.options || [];
      if (!options.some(option => option.id === out[spec.id])) {
        out[spec.id] = (spec.default ?? options[0]?.id) ?? "";
      }
    }
    for (const [key, bounds] of Object.entries(commandBuilder.resource?.limits || ({}))) {
      const fallback = Number((commandBuilder.defaultSelection?.[key] ?? bounds.min) ?? 1);
      const value = Number.parseInt(out[key], 10);
      out[key] = Math.min(bounds.max, Math.max(bounds.min, Number.isFinite(value) ? value : fallback));
    }
    for (const key of ["tp_size", "ulysses_degree", "ring_degree"]) {
      const value = Number.parseInt(out[key], 10);
      out[key] = Number.isFinite(value) && value > 0 ? value : 1;
    }
    out.topology_mode = out.topology_mode === "manual" ? "manual" : "auto";
    return out;
  };
  const placeholderDefaults = schema => {
    const out = {};
    for (const [k, v] of Object.entries(schema || ({}))) out[k] = v.default ?? "";
    return out;
  };
  const [isDark, setIsDark] = useState(false);
  useEffect(() => {
    const check = () => {
      const html = document.documentElement;
      setIsDark(html.classList.contains("dark") || html.getAttribute("data-theme") === "dark" || html.style.colorScheme === "dark");
    };
    check();
    const observer = new MutationObserver(check);
    observer.observe(document.documentElement, {
      attributes: true,
      attributeFilter: ["class", "data-theme", "style"]
    });
    return () => observer.disconnect();
  }, []);
  const STORAGE_KEY = "sglang-deploy-env";
  const [env, setEnv] = useState(() => placeholderDefaults(config.placeholders));
  useEffect(() => {
    try {
      const raw = window.localStorage.getItem(STORAGE_KEY);
      if (raw) {
        const parsed = JSON.parse(raw);
        setEnv({
          ...placeholderDefaults(config.placeholders),
          ...parsed
        });
      }
    } catch {}
  }, []);
  const saveEnv = next => {
    setEnv(next);
    try {
      window.localStorage.setItem(STORAGE_KEY, JSON.stringify(next));
    } catch {}
  };
  const [sel, setSel] = useState(() => initialSelectionFromCells());
  const INTERNAL_HASH_STATE_KEY = "__sglangDeployInternalHash";
  const DEPLOYMENT_COMPONENT_ID = "deployment-configurator";
  useEffect(() => {
    const hydrate = () => {
      const raw = window.location.hash.replace(/^#/, "");
      if (!raw) return;
      const params = new URLSearchParams(raw);
      const initial = initialSelectionFromCells();
      const parsed = {
        ...initial
      };
      let touched = false;
      params.forEach((value, key) => {
        if ((key in parsed)) {
          parsed[key] = value;
          touched = true;
        }
      });
      if (!touched) return;
      setSel(commandBuilder ? normalizeBuilderSelection(parsed) : validateSelection(config.cells, parsed));
      const historyState = window.history.state;
      const isInternalHash = historyState && typeof historyState === "object" && historyState[INTERNAL_HASH_STATE_KEY] === `#${raw}`;
      if (isInternalHash) return;
      const el = document.getElementById(DEPLOYMENT_COMPONENT_ID);
      if (el) el.scrollIntoView({
        behavior: "smooth",
        block: "start"
      });
    };
    hydrate();
    window.addEventListener("hashchange", hydrate);
    return () => window.removeEventListener("hashchange", hydrate);
  }, []);
  useEffect(() => {
    const target = "#" + new URLSearchParams(sel).toString();
    if (window.location.hash !== target) {
      const historyState = window.history.state && typeof window.history.state === "object" ? window.history.state : {};
      window.history.replaceState({
        ...historyState,
        [INTERNAL_HASH_STATE_KEY]: target
      }, "", target);
    }
    window.dispatchEvent(new CustomEvent("sglang-deploy-sel", {
      detail: sel
    }));
  }, [sel]);
  const [modal, setModal] = useState(null);
  useEffect(() => {
    if (modal === null) return;
    const onKey = e => {
      if (e.key === "Escape") setModal(null);
    };
    const prev = document.body.style.overflow;
    document.body.style.overflow = "hidden";
    window.addEventListener("keydown", onKey);
    return () => {
      window.removeEventListener("keydown", onKey);
      document.body.style.overflow = prev;
    };
  }, [modal]);
  const [copied, setCopied] = useState(false);
  const [curlCopied, setCurlCopied] = useState(false);
  const [envDraft, setEnvDraft] = useState(env);
  const [benchConc, setBenchConc] = useState(null);
  const [benchAcc, setBenchAcc] = useState(null);
  const [benchCopied, setBenchCopied] = useState(null);
  const configuredRunModes = typeof config.runModes === "function" ? config.runModes(sel) : config.runModes;
  const runModes = configuredRunModes || ["python", "docker"];
  const [runMode, setRunMode] = useState(runModes[0]);
  const [builderScope, setBuilderScope] = useState("base");
  const [builderServerSetting, setBuilderServerSetting] = useState(null);
  const [builderAdvanced, setBuilderAdvanced] = useState(false);
  const [serveExpanded, setServeExpanded] = useState(false);
  const [requestExpanded, setRequestExpanded] = useState(false);
  const [builderHeadAddress, setBuilderHeadAddress] = useState("<head-node-ip>");
  const [builderNodeRank, setBuilderNodeRank] = useState(0);
  const [blockedNote, setBlockedNote] = useState(null);
  const flashBlockedNote = (dim, reason) => {
    const note = {
      dim,
      reason
    };
    setBlockedNote(note);
    setTimeout(() => setBlockedNote(cur => cur === note ? null : cur), 4000);
  };
  useEffect(() => {
    if (builderNodeRank >= Number(sel.nodes || 1)) setBuilderNodeRank(0);
  }, [sel.nodes, builderNodeRank]);
  const hasRunMode = runModes.includes(runMode);
  const fallbackRunMode = runModes[0];
  const activeRunMode = hasRunMode ? runMode : fallbackRunMode;
  useEffect(() => {
    if (!hasRunMode) setRunMode(fallbackRunMode);
  }, [hasRunMode, fallbackRunMode]);
  useEffect(() => {
    if (modal === "env") setEnvDraft(env);
  }, [modal, env]);
  const [mambaRatio, setMambaRatio] = useState(null);
  useEffect(() => {
    const onRatio = e => setMambaRatio(e.detail && (e.detail.baseRatio || e.detail.ratio) || null);
    window.addEventListener("sglang-k3-mamba-ratio", onRatio);
    return () => window.removeEventListener("sglang-k3-mamba-ratio", onRatio);
  }, []);
  const s = makeStyles(isDark);
  const cell = commandBuilder ? commandBuilder.resolveDeployment(sel) : findCell(config.cells, sel);
  const builderMeta = cell && cell.builder || ({});
  const verifyStatus = cellVerifyStatus(cell, sel);
  const cellWithRatio = (() => {
    if (!cell || !mambaRatio) return cell;
    if (cell.flags.some(f => f.startsWith("--mamba-full-memory-ratio") || f.startsWith("--max-mamba-cache-size"))) return cell;
    const flags = [...cell.flags];
    const line = `--mamba-full-memory-ratio ${mambaRatio}`;
    const i = flags.findIndex(f => f.startsWith("--host"));
    if (i >= 0) flags.splice(i, 0, line); else flags.push(line);
    return {
      ...cell,
      flags
    };
  })();
  const commandEnv = commandBuilder ? {
    ...env,
    NODE_RANK: String(builderNodeRank),
    NODE0_IP: builderHeadAddress || "<head-node-ip>"
  } : env;
  const command = renderCommand(cellWithRatio, sel, commandEnv, activeRunMode);
  const effFlags = cell ? overlayCompose(cell.flags, sel) : [];
  const specAlgoFlag = effFlags.find(f => f.split(/[\s=]/)[0] === "--speculative-algorithm");
  const specMrrFlag = effFlags.find(f => f.split(/[\s=]/)[0] === "--max-running-requests");
  const mtpHint = !!specAlgoFlag && !specMrrFlag;
  const specPinnedHint = !!specAlgoFlag && !!specMrrFlag;
  const specMrrValue = specMrrFlag ? specMrrFlag.split(/[\s=]/).filter(Boolean)[1] || "" : "";
  const SPEC_ALGO_LABEL = {
    EAGLE: "MTP",
    EAGLE3: "MTP",
    FROZEN_KV_MTP: "MTP",
    DSPARK: "DSpark",
    DFLASH: "DFlash",
    NGRAM: "N-gram",
    STANDALONE: "standalone draft"
  };
  const specAlgoName = (() => {
    if (!specAlgoFlag) return "MTP";
    const v = specAlgoFlag.split(/[\s=]/).filter(Boolean)[1] || "";
    return SPEC_ALGO_LABEL[v.toUpperCase()] || v || "MTP";
  })();
  const renderWarn = text => {
    const out = [];
    const re = /\[([^\]]+)\]\(#([^)]+)\)/g;
    let last = 0;
    for (let m; m = re.exec(text); last = m.index + m[0].length) {
      if (m.index > last) out.push(text.slice(last, m.index));
      const anchor = m[2];
      out.push(<button key={m.index} type="button" onClick={() => {
        const el = document.getElementById(anchor);
        if (el) el.scrollIntoView({
          behavior: "smooth",
          block: "start"
        });
      }} style={{
        background: "transparent",
        border: "none",
        padding: 0,
        color: isDark ? "#FDBA74" : "#C2410C",
        cursor: "pointer",
        font: "inherit",
        fontWeight: 600,
        textDecoration: "underline",
        textUnderlineOffset: "2px"
      }}>
          {m[1]}
        </button>);
    }
    if (last < text.length) out.push(text.slice(last));
    return out;
  };
  const modelName = resolveModelName(sel);
  const curlTemplate = typeof config.curl === "function" ? config.curl(sel, cell) : config.curl;
  const curlText = interpolate(curlTemplate || "", env, modelName);
  const hwGroups = buildHardwareGroups();
  const benchEntry = benchmarks ? findBenchmark(benchmarks, sel) : null;
  const isOverlayDim = dim => overlayDimSpecs.some(d => d.id === dim);
  const findOption = (dim, value) => {
    const spec = [...matchDimSpecs, ...overlayDimSpecs].find(d => d.id === dim);
    return spec && (spec.options || []).find(o => o.id === value);
  };
  const isEnabled = (dim, value) => {
    const opt = findOption(dim, value);
    if (opt && optionDisabled(opt, sel)) return false;
    if (commandBuilder && dim === "hw") return true;
    return isOverlayDim(dim) || isOptionAvailable(config.cells || [], sel, dim, value);
  };
  const reseatHiddenPicks = next => {
    let out = next;
    for (const spec of [...matchDimSpecs, ...overlayDimSpecs]) {
      const opts = visibleOptions(spec, out).filter(o => !optionDisabled(o, out));
      if (!opts.length) continue;
      if (!opts.some(o => o.id === out[spec.id])) {
        out = {
          ...out,
          [spec.id]: opts[0].id
        };
      }
    }
    return out;
  };
  const recommendedBuilderRecipe = hw => {
    const recipes = commandBuilder.resource?.verifiedRecipes || [];
    return recipes.find(entry => entry.hw === hw && entry.default) || recipes.find(entry => entry.hw === hw);
  };
  const handleSelect = (dim, value) => {
    if (commandBuilder) {
      setSel(prev => {
        let next = {
          ...prev,
          [dim]: value
        };
        if (dim === "hw") {
          const currentRecipe = recommendedBuilderRecipe(prev.hw);
          const nextRecipe = recommendedBuilderRecipe(value);
          const resourcesFollowPlatformDefault = !!currentRecipe && Number(prev.nodes) === Number(currentRecipe.nodes) && Number(prev.gpus_per_node) === Number(currentRecipe.gpus_per_node);
          next = {
            ...next,
            nodes: resourcesFollowPlatformDefault ? nextRecipe?.nodes ?? next.nodes : next.nodes,
            gpus_per_node: resourcesFollowPlatformDefault ? nextRecipe?.gpus_per_node ?? next.gpus_per_node : next.gpus_per_node,
            topology_mode: "auto",
            tp_size: resourcesFollowPlatformDefault ? nextRecipe?.tp_size ?? 1 : next.tp_size,
            ulysses_degree: resourcesFollowPlatformDefault ? nextRecipe?.ulysses_degree ?? 1 : next.ulysses_degree,
            ring_degree: resourcesFollowPlatformDefault ? nextRecipe?.ring_degree ?? 1 : next.ring_degree,
            placement: resourcesFollowPlatformDefault ? nextRecipe?.placement || "auto" : next.placement,
            encoder: resourcesFollowPlatformDefault ? nextRecipe?.encoder || "auto" : next.encoder
          };
        }
        return reseatHiddenPicks(normalizeBuilderSelection(next));
      });
      return;
    }
    setSel(prev => reseatHiddenPicks(isOverlayDim(dim) ? {
      ...prev,
      [dim]: value
    } : snapToValidCell(config.cells, prev, dim, value)));
  };
  const commitBuilderNumber = (event, currentValue, bounds, commit) => {
    const parsed = Number(event.currentTarget.value);
    if (!Number.isInteger(parsed)) {
      event.currentTarget.value = String(currentValue);
      return;
    }
    const value = Math.min(bounds.max, Math.max(bounds.min, parsed));
    event.currentTarget.value = String(value);
    commit(value);
  };
  const renderBuilderNumberInput = ({identity, value, min, max, label, onCommit}) => <input key={identity} type="number" inputMode="numeric" min={min} max={max} step="1" defaultValue={value} aria-label={label} onFocus={event => event.currentTarget.select()} onBlur={event => commitBuilderNumber(event, value, {
    min,
    max
  }, onCommit)} onKeyDown={event => {
    if (event.key === "Enter") event.currentTarget.blur();
  }} />;
  const updateBuilderResource = (key, delta) => {
    if (!commandBuilder) return;
    const bounds = commandBuilder.resource?.limits?.[key] || ({
      min: 1,
      max: 8
    });
    setSel(prev => {
      const value = Math.min(bounds.max, Math.max(bounds.min, Number(prev[key]) + delta));
      return normalizeBuilderSelection({
        ...prev,
        [key]: value,
        topology_mode: "auto"
      });
    });
  };
  const setBuilderResource = (key, rawValue) => {
    if (!commandBuilder) return;
    const value = Number.parseInt(rawValue, 10);
    if (!Number.isFinite(value)) return;
    const bounds = commandBuilder.resource?.limits?.[key] || ({
      min: 1,
      max: 8
    });
    setSel(prev => normalizeBuilderSelection({
      ...prev,
      [key]: Math.min(bounds.max, Math.max(bounds.min, value)),
      topology_mode: "auto"
    }));
  };
  const editBuilderTopology = (key, value) => {
    if (!commandBuilder) return;
    setSel(prev => normalizeBuilderSelection({
      ...prev,
      topology_mode: "manual",
      [key]: Number.parseInt(value, 10) || 1
    }));
  };
  const handleCopy = () => {
    navigator.clipboard.writeText(command);
    setCopied(true);
    setTimeout(() => setCopied(false), 1200);
  };
  const copyCurl = () => {
    navigator.clipboard.writeText(curlText);
    setCurlCopied(true);
    setTimeout(() => setCurlCopied(false), 1200);
  };
  const copyBench = (key, text) => {
    navigator.clipboard.writeText(text);
    setBenchCopied(key);
    setTimeout(() => setBenchCopied(null), 1200);
  };
  const placeholderGroups = (() => {
    const out = {
      command: [],
      curl: []
    };
    for (const [key, meta] of Object.entries(config.placeholders || ({}))) {
      (out[meta.target] || (out[meta.target] = [])).push({
        key,
        ...meta
      });
    }
    return out;
  })();
  const renderButton = (item, dim, selectedId) => {
    const checked = selectedId === item.id;
    const disabled = !isEnabled(dim, item.id);
    return <label key={item.id} className="sg-command-visualizer-choice" role="radio" aria-checked={checked} aria-disabled={disabled} tabIndex={disabled ? -1 : 0} style={{
      ...s.labelBase,
      ...checked ? s.checked : {},
      ...disabled ? s.disabled : {}
    }} title={disabled ? (typeof item.disableReason === "function" ? item.disableReason(sel) : item.disableReason) || "Not supported for current selection" : ""} onClick={e => {
      if (disabled) {
        e.preventDefault();
        return;
      }
      handleSelect(dim, item.id);
    }} onKeyDown={e => {
      if (disabled || e.key !== "Enter" && e.key !== " ") return;
      e.preventDefault();
      handleSelect(dim, item.id);
    }}>
        <input type="radio" checked={checked} disabled={disabled} readOnly style={{
      display: "none"
    }} />
        <span>{item.label}</span>
        {item.subtitle && <small style={{
      ...s.subtitle,
      color: checked ? "rgba(255,255,255,0.85)" : "inherit"
    }}>
            {item.subtitle}
          </small>}
      </label>;
  };
  const renderFlatSection = (title, options, dim, selectedId) => <div style={s.card}>
      <div style={s.title}>{title}</div>
      <div style={s.itemsGrid(options.length)}>
        {options.map(item => renderButton(item, dim, selectedId))}
      </div>
    </div>;
  const maxHwCols = Math.max(...hwGroups.map(x => x.items.length));
  if (commandBuilder) {
    const scopeLabel = {
      base: "Setup",
      serve: "Server",
      request: "Request"
    };
    const scopedDims = scope => overlayDimSpecs.filter(dim => {
      if ((dim.scope || "base") !== scope) return false;
      if (dim.kind === "number") {
        return typeof dim.showWhen !== "function" || dim.showWhen(sel);
      }
      return rowVisible(dim, sel);
    });
    const baseDims = scopedDims("base");
    const serveDims = scopedDims("serve");
    const requestDims = scopedDims("request");
    const errors = builderMeta.errors || [];
    const warnings = builderMeta.warnings || [];
    const invalid = errors.length > 0;
    const totalGpus = Number(sel.nodes) * Number(sel.gpus_per_node);
    const topology = builderMeta.topology || ({});
    const verification = builderMeta.verification || ({});
    const scopeIsVerified = scope => scopedDims(scope).every(dim => {
      const option = (dim.options || []).find(entry => entry.id === sel[dim.id]);
      if (option && optionSoft(option, sel)) return false;
      const predicate = option?.verifiedWhen ?? dim.verifiedWhen;
      return typeof predicate === "function" ? !!predicate(sel) : predicate !== false;
    });
    const serveStatus = invalid ? "error" : scopeIsVerified("serve") ? verification.serve || verifyStatus : "unverified";
    const requestStatus = invalid ? "error" : scopeIsVerified("request") ? verification.request || verifyStatus : "unverified";
    const statusText = status => ({
      verified: "Verified",
      unverified: "Unverified",
      "in-progress": "Verification in progress",
      error: "Invalid configuration"
    })[status] || "Unverified";
    const activeServerSetting = serveDims.find(dim => dim.id === builderServerSetting) || serveDims[0];
    const selectedOption = dim => (dim.options || []).find(option => option.id === sel[dim.id]);
    const effectiveSetting = dim => builderMeta.resolvedSettings?.[dim.id] || selectedOption(dim)?.label || sel[dim.id] || "—";
    const recommendedRecipe = recommendedBuilderRecipe(sel.hw);
    const recommendedInUse = !!recommendedRecipe && Number(sel.nodes) === recommendedRecipe.nodes && Number(sel.gpus_per_node) === recommendedRecipe.gpus_per_node && sel.topology_mode === "auto" && ["auto", recommendedRecipe.placement].includes(sel.placement) && sel.attention === "platform" && sel.precision === "native" && ["auto", recommendedRecipe.encoder].includes(sel.encoder) && sel.execution === "eager";
    const restoreRecommendedRecipe = () => {
      if (!recommendedRecipe) return;
      setSel(prev => reseatHiddenPicks(normalizeBuilderSelection({
        ...prev,
        nodes: recommendedRecipe.nodes,
        gpus_per_node: recommendedRecipe.gpus_per_node,
        topology_mode: "auto",
        tp_size: recommendedRecipe.tp_size,
        ulysses_degree: recommendedRecipe.ulysses_degree,
        ring_degree: recommendedRecipe.ring_degree,
        placement: recommendedRecipe.placement || "auto",
        attention: "platform",
        precision: "native",
        encoder: recommendedRecipe.encoder || "auto",
        execution: "eager"
      })));
    };
    const renderBuilderChoice = (item, dim) => {
      const checked = sel[dim.id] === item.id;
      const disabled = !isEnabled(dim.id, item.id);
      const soft = !disabled && optionSoft(item, sel);
      const reason = disabled ? item.disableReason || "Not available for this configuration" : soft ? item.softReason || "Runs, but this combination is not a verified recipe yet." : "";
      return <button key={item.id} type="button" className="sgd-builder-choice" data-selected={checked ? "true" : "false"} data-blocked={disabled ? "true" : undefined} data-soft={soft ? "true" : undefined} aria-disabled={disabled} aria-pressed={checked} title={reason} onClick={() => {
        if (disabled) {
          flashBlockedNote(dim.id, reason);
          return;
        }
        handleSelect(dim.id, item.id);
      }}>
          <span className="sgd-builder-choice-dot" aria-hidden="true" />
          <span>{item.label}</span>
          {item.subtitle && <small>{item.subtitle}</small>}
        </button>;
    };
    const renderBuilderDimension = dim => <section className="sgd-builder-section" key={dim.id}>
        <div className="sgd-builder-section-heading">
          <span>{dim.title}</span>
          {dim.description && <small>{dim.description}</small>}
        </div>
        <div className="sgd-builder-choice-grid" data-density={(dim.options || []).length > 5 ? "compact" : "normal"}>
          {visibleOptions(dim, sel).map(option => renderBuilderChoice(option, dim))}
        </div>
        {blockedNote && blockedNote.dim === dim.id && <p className="sgd-builder-blocked-note" role="status">{blockedNote.reason}</p>}
      </section>;
    const renderStepper = (key, label, detail) => {
      const bounds = commandBuilder.resource?.limits?.[key] || ({
        min: 1,
        max: 8
      });
      return <div className="sgd-builder-stepper-field">
          <div>
            <span>{label}</span>
            {detail && <small>{detail}</small>}
          </div>
          <div className="sgd-builder-stepper" aria-label={label}>
            <button type="button" aria-label={`Decrease ${label}`} disabled={Number(sel[key]) <= bounds.min} onClick={() => updateBuilderResource(key, -1)}>−</button>
            {renderBuilderNumberInput({
        identity: `${key}-${sel[key]}`,
        value: sel[key],
        min: bounds.min,
        max: bounds.max,
        label,
        onCommit: value => setBuilderResource(key, value)
      })}
            <button type="button" aria-label={`Increase ${label}`} disabled={Number(sel[key]) >= bounds.max} onClick={() => updateBuilderResource(key, 1)}>+</button>
          </div>
        </div>;
    };
    const renderBaseScope = () => <div className="sgd-builder-scope-panel" data-scope="base">
        {recommendedRecipe && <section className="sgd-builder-recipe">
            <div>
              {}
              <span>{recommendedRecipe.unverified ? "Derived recipe" : "Verified recipe"} · {sel.hw.toUpperCase()}</span>
              <strong>
                {[`${recommendedRecipe.nodes * recommendedRecipe.gpus_per_node} GPUs`, recommendedRecipe.tp_size > 1 && `TP ${recommendedRecipe.tp_size}`, `Ulysses ${recommendedRecipe.ulysses_degree}`, recommendedRecipe.ring_degree > 1 && `Ring ${recommendedRecipe.ring_degree}`, ({
      resident: "Resident",
      fsdp: "FSDP",
      offload: "Layerwise offload"
    })[recommendedRecipe.placement]].filter(Boolean).join(" · ")}
              </strong>
            </div>
            <div>
              {renderStatus(recommendedRecipe.unverified ? "unverified" : "verified")}
              {recommendedInUse ? <small>In use</small> : <button type="button" className="sgd-builder-text-action" onClick={restoreRecommendedRecipe}>{recommendedRecipe.unverified ? "Use derived recipe" : "Use verified recipe"}</button>}
            </div>
          </section>}
        <section className="sgd-builder-section">
          <div className="sgd-builder-section-heading"><span>Hardware</span></div>
          <div className="sgd-builder-hardware-grid">
            {hwGroups.flatMap(group => group.items).map(item => {
      const selected = sel.hw === item.id;
      return <button key={item.id} type="button" className="sgd-builder-hardware" data-selected={selected ? "true" : "false"} aria-pressed={selected} onClick={() => handleSelect("hw", item.id)}>
                  <span className="sgd-builder-choice-dot" aria-hidden="true" />
                  <strong>{item.label}</strong>
                  <small>{item.subtitle}</small>
                </button>;
    })}
          </div>
        </section>

        {}
        <section className="sgd-builder-section">
          <div className="sgd-builder-section-heading">
            <span>Resources</span>
            <small>{builderMeta.topologySummary || "No valid topology"}</small>
          </div>
          <div className="sgd-builder-resource-grid">
            {renderStepper("nodes", "Nodes")}
            {renderStepper("gpus_per_node", "GPUs / node")}
          </div>
          {Number(sel.nodes) > 1 && <p className="sgd-builder-resource-summary">
              {sel.nodes} nodes × {sel.gpus_per_node} {sel.hw.toUpperCase()} = {totalGpus} GPUs
            </p>}
          <button type="button" className="sgd-builder-text-action sgd-builder-topology-toggle" aria-expanded={builderAdvanced} onClick={() => setBuilderAdvanced(open => !open)}>
            Advanced topology <span aria-hidden="true">{builderAdvanced ? "↗" : "↘"}</span>
          </button>
          {builderAdvanced && <div className="sgd-builder-advanced">
              <p>Auto uses an exact verified recipe when one exists; manual values are allowed when the model constraints remain valid.</p>
              <div className="sgd-builder-topology-inputs">
                {[["tp_size", "Tensor parallel", [1, 2, 4, 8]], ["ulysses_degree", "Ulysses", [1, 2, 4, 8, 16]], ["ring_degree", "Ring", [1, 2, 4, 8]]].map(([key, label, values]) => <label key={key}>
                    <span>{label}</span>
                    <select value={sel.topology_mode === "manual" ? sel[key] : topology[key] || 1} onChange={event => editBuilderTopology(key, event.target.value)}>
                      {values.map(value => <option value={value} key={value}>{value}</option>)}
                    </select>
                  </label>)}
              </div>
              <button type="button" className="sgd-builder-text-action" disabled={sel.topology_mode === "auto"} onClick={() => setSel(prev => normalizeBuilderSelection({
      ...prev,
      topology_mode: "auto"
    }))}>Use automatic topology</button>
            </div>}
          {(errors.length > 0 || warnings.length > 0) && <div className="sgd-builder-messages" data-state={errors.length ? "error" : "warning"}>
              {(errors.length ? errors : warnings).map((message, index) => <p key={index}>{message}</p>)}
            </div>}
        </section>

        {baseDims.map(renderBuilderDimension)}
      </div>;
    const renderSettingEditor = (dim, className = "", direct = false) => {
      if (!dim) return null;
      const options = visibleOptions(dim, sel);
      const currentOption = selectedOption(dim);
      return <section className={`sgd-builder-context ${className}`} aria-live={direct ? undefined : "polite"}>
          <div className="sgd-builder-context-heading">
            <div>
              <span>{direct ? dim.title : `${dim.title} options`}</span>
              {dim.description && <p>{dim.description}</p>}
            </div>
            {dim.quality && <small>{dim.quality}</small>}
          </div>
          {dim.kind === "number" ? <div className="sgd-builder-request-stepper">
              <button type="button" aria-label={`Decrease ${dim.title}`} disabled={Number(sel[dim.id]) <= dim.min} onClick={() => setSel(prev => ({
        ...prev,
        [dim.id]: Math.max(dim.min, Number(prev[dim.id]) - 1)
      }))}>−</button>
              {renderBuilderNumberInput({
        identity: `${dim.id}-${sel[dim.id]}`,
        value: sel[dim.id],
        min: dim.min,
        max: dim.max,
        label: dim.title,
        onCommit: value => setSel(prev => ({
          ...prev,
          [dim.id]: value
        }))
      })}
              <button type="button" aria-label={`Increase ${dim.title}`} disabled={Number(sel[dim.id]) >= dim.max} onClick={() => setSel(prev => ({
        ...prev,
        [dim.id]: Math.min(dim.max, Number(prev[dim.id]) + 1)
      }))}>+</button>
              <span>{dim.unit || "outputs"}</span>
            </div> : <div className="sgd-builder-context-options">
              {options.map(option => renderBuilderChoice(option, dim))}
            </div>}
          {blockedNote && blockedNote.dim === dim.id && <p className="sgd-builder-blocked-note" role="status">{blockedNote.reason}</p>}
          {(currentOption?.description || dim.learnMore) && <div className="sgd-builder-context-note">
              {currentOption?.description && <p>{currentOption.description}</p>}
              {}
              {dim.learnMore && <a href={dim.learnMore}>
                  <svg width="12" height="12" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" aria-hidden="true"><line x1="4" y1="6" x2="20" y2="6" /><line x1="4" y1="12" x2="16" y2="12" /><line x1="4" y1="18" x2="11" y2="18" /></svg>
                  Learn more
                </a>}
              {dim.docsHref && <a href={dim.docsHref}>
                  <svg width="12" height="12" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round" aria-hidden="true"><path d="M4 19.5A2.5 2.5 0 0 1 6.5 17H20" /><path d="M6.5 2H20v20H6.5A2.5 2.5 0 0 1 4 19.5v-15A2.5 2.5 0 0 1 6.5 2z" /></svg>
                  SGLang docs
                </a>}
            </div>}
        </section>;
    };
    const renderServerScope = () => <div className="sgd-builder-scope-panel" data-scope="serve">
        <div className="sgd-builder-setting-layout">
          <div className="sgd-builder-setting-list">
            {serveDims.map(dim => {
      const isActive = dim.id === activeServerSetting?.id;
      const option = selectedOption(dim);
      const recommended = typeof option?.recommendedWhen === "function" ? option.recommendedWhen(sel) : !!option?.recommended;
      return <div className="sgd-builder-setting-item" key={dim.id}>
                  <button type="button" className="sgd-builder-setting-row" data-active={isActive ? "true" : "false"} aria-expanded={isActive} onClick={() => setBuilderServerSetting(dim.id)}>
                    <span>{dim.title}</span>
                    <strong>{effectiveSetting(dim)}</strong>
                    {recommended && <small>Recommended</small>}
                    <span aria-hidden="true">{isActive ? "⌄" : "›"}</span>
                  </button>
                  {isActive && renderSettingEditor(dim, "sgd-builder-context--inline")}
                </div>;
    })}
          </div>
          {renderSettingEditor(activeServerSetting, "sgd-builder-context--rail")}
        </div>
      </div>;
    const renderRequestScope = () => <div className="sgd-builder-scope-panel sgd-builder-request-direct" data-scope="request">
        {requestDims.map(dim => <div className="sgd-builder-request-setting" key={dim.id}>
            {renderSettingEditor(dim, "", true)}
          </div>)}
      </div>;
    const renderScopeControls = () => {
      if (builderScope === "base") return renderBaseScope();
      if (builderScope === "serve") return renderServerScope();
      return renderRequestScope();
    };
    const renderStatus = status => <span className="sgd-builder-status" data-status={status}>
        <span aria-hidden="true" />{statusText(status)}
      </span>;
    const renderOutputCard = type => {
      const serve = type === "serve";
      const text = serve ? command : curlText;
      const canExpand = text.split("\n").length > 9;
      const expanded = serve ? serveExpanded : requestExpanded;
      const setExpanded = serve ? setServeExpanded : setRequestExpanded;
      const status = serve ? serveStatus : requestStatus;
      const emphasized = builderScope === "base" || builderScope === type;
      return <section className="sgd-builder-output" data-output={type} data-emphasized={emphasized ? "true" : "false"}>
          <header>
            <div className="sgd-builder-output-index">{serve ? "1" : "2"}</div>
            <div className="sgd-builder-output-title">
              <strong>{serve ? "Serve" : "Request"}</strong>
              <span>
                {serve ? `${sel.hw.toUpperCase()} · ${activeRunMode === "docker" ? "Docker" : "Python"}` : "cURL"}
              </span>
            </div>
            {renderStatus(status)}
          </header>
          {serve && runModes.length > 1 && <div className="sgd-builder-output-tabs" role="tablist" aria-label="Serve command format">
              {runModes.map(mode => <button type="button" role="tab" aria-selected={activeRunMode === mode} data-selected={activeRunMode === mode ? "true" : "false"} key={mode} onClick={() => setRunMode(mode)}>{mode === "docker" ? "Docker" : "Python"}</button>)}
            </div>}
          {serve && Number(sel.nodes) > 1 && <div className="sgd-builder-node-fields">
              <label>
                <span>Head address</span>
                <input value={builderHeadAddress} onChange={event => setBuilderHeadAddress(event.target.value)} />
              </label>
              <label>
                <span>Node rank</span>
                {renderBuilderNumberInput({
        identity: `node-rank-${builderNodeRank}-${sel.nodes}`,
        value: builderNodeRank,
        min: 0,
        max: Number(sel.nodes) - 1,
        label: "Node rank",
        onCommit: setBuilderNodeRank
      })}
              </label>
            </div>}
          <div className="sgd-builder-code">
            <pre className={expanded ? "is-expanded" : ""}><code>{text}</code></pre>
            <button type="button" className="sgd-builder-copy" disabled={invalid} aria-label={(serve ? copied : curlCopied) ? "Copied" : "Copy command"} data-copied={(serve ? copied : curlCopied) ? "true" : undefined} onClick={serve ? handleCopy : copyCurl}>
              <svg className="sgd-builder-copy-glyph" width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="1.9" strokeLinecap="round" strokeLinejoin="round" aria-hidden="true"><rect x="9" y="9" width="12" height="12" rx="2.5" /><path d="M15 5v-.25A2.75 2.75 0 0 0 12.25 2h-7.5A2.75 2.75 0 0 0 2 4.75v7.5A2.75 2.75 0 0 0 4.75 15H5" /></svg>
              <svg className="sgd-builder-copy-check" width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2.2" strokeLinecap="round" strokeLinejoin="round" aria-hidden="true"><path d="M20 6 9 17l-5-5" /></svg>
            </button>
          </div>
          {invalid && <div className="sgd-builder-output-error">{errors[0]}</div>}
          <footer>
            {canExpand && <button type="button" className="sgd-builder-text-action" onClick={() => setExpanded(!expanded)}>
                {expanded ? "Collapse" : "Expand"}
              </button>}
            <div>
              <button type="button" className="sgd-builder-text-action" onClick={() => setModal("env")}>Variables</button>
            </div>
          </footer>
        </section>;
    };
    return <section id={DEPLOYMENT_COMPONENT_ID} className="not-prose sg-command-visualizer sgd-command-builder" style={{
      scrollMarginTop: "104px"
    }} aria-label={`${config.modelName} command builder`}>
        <nav className="sgd-builder-scope-tabs" role="tablist" aria-label="Command builder scope">
          {["base", "serve", "request"].map(scope => <button type="button" role="tab" key={scope} aria-label={scopeLabel[scope]} aria-selected={builderScope === scope} aria-controls={`${DEPLOYMENT_COMPONENT_ID}-controls`} data-active={builderScope === scope ? "true" : "false"} onClick={() => setBuilderScope(scope)}>
              {scopeLabel[scope]}
            </button>)}
        </nav>

        <div className="sgd-builder-main" data-scope={builderScope}>
          <div id={`${DEPLOYMENT_COMPONENT_ID}-controls`} className="sgd-builder-controls" role="tabpanel" aria-label={`${scopeLabel[builderScope]} settings`}>
            {renderScopeControls()}
          </div>
          <div className="sgd-builder-output-rail">
            {builderScope !== "request" && renderOutputCard("serve")}
            {builderScope !== "serve" && renderOutputCard("request")}
          </div>
        </div>

        {modal === "env" && <div style={s.modalBackdrop} onClick={() => setModal(null)}>
            <div style={s.modalBox} onClick={event => event.stopPropagation()}>
              <div style={s.modalHeader}>
                <div style={s.modalTitle}>Command variables</div>
                <button style={s.modalCloseBtn} onClick={() => setModal(null)} aria-label="Close">×</button>
              </div>
              {["command", "curl"].map(target => placeholderGroups[target].length > 0 && <div key={target}>
                  <div style={s.sectionHeading}>{target === "command" ? "Serve" : "Request"}</div>
                  {placeholderGroups[target].map(({key, label}) => <div key={key} style={s.formField}>
                      <label style={s.formLabel}>{label}</label>
                      <input style={s.formInput} value={envDraft[key] ?? ""} onChange={event => setEnvDraft({
      ...envDraft,
      [key]: event.target.value
    })} />
                    </div>)}
                </div>)}
              <div style={{
      display: "flex",
      justifyContent: "flex-end",
      gap: 8,
      marginTop: 16
    }}>
                <button style={{
      ...s.iconButton,
      padding: "6px 14px"
    }} onClick={() => setModal(null)}>Cancel</button>
                <button style={s.primaryBtn} onClick={() => {
      saveEnv(envDraft);
      setModal(null);
    }}>Save</button>
              </div>
            </div>
          </div>}
      </section>;
  }
  return <div id={DEPLOYMENT_COMPONENT_ID} style={{
    ...s.container,
    scrollMarginTop: "104px"
  }} className="not-prose sg-command-visualizer">
      {}
      <div style={s.cardColumn}>
        <div style={{
    ...s.title,
    marginBottom: "2px"
  }}>Hardware Platform</div>
        {hwGroups.map(g => <div key={g.label || "hardware"} style={s.vendorRow}>
            {g.label && <div style={s.vendorLabel}>{g.label}</div>}
            <div style={s.itemsGrid(maxHwCols)}>
              {g.items.map(item => renderButton(item, "hw", sel.hw))}
              {Array.from({
    length: maxHwCols - g.items.length
  }).map((_, i) => <div key={`pad-${i}`} />)}
            </div>
          </div>)}
      </div>

      {matchDimSpecs.filter(d => rowVisible(d, sel)).map(d => <div key={d.id}>
            {renderFlatSection(d.title, visibleOptions(d, sel), d.id, sel[d.id])}
          </div>)}
      {overlayDimSpecs.filter(d => rowVisible(d, sel)).map(d => <div key={d.id}>
            {renderFlatSection(d.title, visibleOptions(d, sel), d.id, sel[d.id])}
          </div>)}

      {}
      <div style={s.card}>
        <div style={s.title}>Command:</div>
        <div style={s.commandWrap}>
          {cell && cell.redirect ? cell.warn && <div style={s.mtpWarn}>⚠️ {renderWarn(cell.warn)}</div> : <>
            <div style={s.commandHeader}>
              <div style={s.headerLeft}>
                <div style={s.badge(verifyStatus)}>
                  <span style={s.badgeDot(verifyStatus)} />
                  {VERIFY_LABEL[verifyStatus]}
                </div>
                <div style={s.runModeWrap} role="tablist" aria-label="Output format">
                  {runModes.map((mode, index) => <span key={mode} className="sg-command-visualizer-tab" style={{
    ...index === runModes.length - 1 ? s.runModeChipLast(activeRunMode === mode) : s.runModeChip(activeRunMode === mode),
    ...runModes.length === 1 ? {
      borderRadius: 7
    } : {}
  }} onClick={() => setRunMode(mode)} onKeyDown={e => {
    if (e.key !== "Enter" && e.key !== " ") return;
    e.preventDefault();
    setRunMode(mode);
  }} role="tab" tabIndex={0} aria-selected={activeRunMode === mode}>
                      {mode === "docker" ? "Docker" : "Python"}
                    </span>)}
                </div>
              </div>
              <div style={s.iconRow}>
                <button style={s.iconButton} onClick={handleCopy}>
                  {copied ? "✓ Copied" : "⧉ Copy"}
                </button>
                <button style={s.iconButton} onClick={() => setModal("curl")}>$ cURL</button>
                <button style={s.iconButton} onClick={() => setModal("env")}>⚙ Env</button>
              </div>
            </div>
            <pre style={s.commandPre}>{command}</pre>
            {cell && cell.warn && <div style={s.mtpWarn}>⚠️ {renderWarn(cell.warn)}</div>}
            {mtpHint && <div style={s.mtpWarn}>
                ⚠️ Speculative decoding ({specAlgoName}) is on — SGLang resets <code>--max-running-requests</code> to <strong>48</strong> when it isn't set. Add <code>--max-running-requests &lt;N&gt;</code> sized for your target concurrency.
              </div>}
            {specPinnedHint && <div style={s.mtpWarn}>
                ℹ️ Speculative decoding ({specAlgoName}) is on and this recipe pins <code>--max-running-requests</code> to <strong>{specMrrValue}</strong>. Adjust it to match your target concurrency — if you remove the flag, SGLang falls back to <strong>48</strong>.
              </div>}
          </>}
        </div>
      </div>

      {}
      {benchmarks && cell && renderBenchmarkCard(benchEntry)}

      {}
      {config.showPlaygroundLink !== false && <div style={{
    padding: "6px 12px",
    fontSize: "12px",
    color: isDark ? "#9ca3af" : "#6b7280",
    display: "flex",
    alignItems: "center",
    gap: "6px"
  }}>
          <span>Need to go beyond the verified matrix?</span>
          <button type="button" onClick={() => {
    const el = document.getElementById("playground");
    if (el) el.scrollIntoView({
      behavior: "smooth",
      block: "start"
    });
  }} style={{
    background: "transparent",
    border: "none",
    padding: 0,
    color: isDark ? "#FDBA74" : "#C2410C",
    cursor: "pointer",
    fontSize: "12px",
    fontWeight: 600,
    textDecoration: "underline",
    textUnderlineOffset: "2px"
  }}>
            Open the Playground →
          </button>
        </div>}

      {}
      {modal === "curl" && <div style={s.modalBackdrop} onClick={() => setModal(null)}>
          <div style={s.modalBox} onClick={e => e.stopPropagation()}>
            <div style={s.modalHeader}>
              <div style={s.modalTitle}>cURL example</div>
              <button style={s.modalCloseBtn} onClick={() => setModal(null)} aria-label="Close">×</button>
            </div>
            <div style={s.commandWrap}>
              <div style={s.commandHeader}>
                <div style={{
    fontSize: 11,
    opacity: 0.7
  }}>
                  Model: <code>{modelName || "(unresolved)"}</code>
                </div>
                <button style={s.iconButton} onClick={copyCurl}>
                  {curlCopied ? "✓ Copied" : "⧉ Copy"}
                </button>
              </div>
              <pre style={s.commandPre}>{curlText}</pre>
            </div>
            <p style={{
    fontSize: 11,
    opacity: 0.7,
    marginTop: 8
  }}>
              Edit <code>CURL_HOST</code> / <code>CURL_PORT</code> in the Env panel.
            </p>
          </div>
        </div>}

      {}
      {modal === "env" && <div style={s.modalBackdrop} onClick={() => setModal(null)}>
          <div style={s.modalBox} onClick={e => e.stopPropagation()}>
            <div style={s.modalHeader}>
              <div style={s.modalTitle}>Env / placeholder values</div>
              <button style={s.modalCloseBtn} onClick={() => setModal(null)} aria-label="Close">×</button>
            </div>
            {placeholderGroups.curl.length > 0 && <div>
                <div style={s.sectionHeading}>cURL placeholders</div>
                {placeholderGroups.curl.map(({key, label}) => <div key={key} style={s.formField}>
                    <label style={s.formLabel}>
                      {label} <code style={{
    opacity: 0.6
  }}>{`{{${key}}}`}</code>
                    </label>
                    <input style={s.formInput} value={envDraft[key] ?? ""} onChange={e => setEnvDraft({
    ...envDraft,
    [key]: e.target.value
  })} />
                  </div>)}
              </div>}
            {placeholderGroups.command.length > 0 && <div>
                <div style={s.sectionHeading}>Command placeholders</div>
                {placeholderGroups.command.map(({key, label}) => <div key={key} style={s.formField}>
                    <label style={s.formLabel}>
                      {label} <code style={{
    opacity: 0.6
  }}>{`{{${key}}}`}</code>
                    </label>
                    <input style={s.formInput} value={envDraft[key] ?? ""} onChange={e => setEnvDraft({
    ...envDraft,
    [key]: e.target.value
  })} />
                  </div>)}
              </div>}
            <div style={{
    display: "flex",
    justifyContent: "flex-end",
    gap: 8,
    marginTop: 16
  }}>
              <button style={{
    ...s.iconButton,
    padding: "6px 14px"
  }} onClick={() => setModal(null)}>Cancel</button>
              <button style={s.primaryBtn} onClick={() => {
    saveEnv(envDraft);
    setModal(null);
  }}>Save</button>
            </div>
            <p style={{
    fontSize: 11,
    opacity: 0.7,
    marginTop: 10
  }}>
              Values persist in localStorage and are reused the next time you visit any cookbook.
            </p>
          </div>
        </div>}

      {}
      {modal === "bench" && benchEntry && (() => {
    const bc = buildBenchCommands(benchEntry, sel);
    if (!bc) return null;
    const selSummary = [sel.hw && sel.hw.toUpperCase(), sel.variant, sel.quant && sel.quant.toUpperCase(), sel.strategy, sel.kvDsaPair, sel.nodes].filter(part => part !== undefined && part !== null && part !== "").join(" · ");
    let selConc = null;
    let speedCmd = null;
    if (bc.speed) {
      selConc = bc.speed.concurrencies.includes(benchConc) ? benchConc : bc.speed.concurrencies[0];
      const w = bc.speed.workload;
      speedCmd = interpolate(bc.speed.template, {
        ...env,
        DATASET: w.dataset,
        ISL: w.isl,
        OSL: w.osl,
        MAX_CONCURRENCY: selConc,
        NUM_PROMPTS: bc.speed.numPromptsOf(selConc)
      }, modelName);
    }
    let selAcc = null;
    let accCmd = null;
    if (bc.accuracy.length > 0) {
      selAcc = bc.accuracy.find(a => a.key === benchAcc) || bc.accuracy[0];
      accCmd = interpolate(selAcc.template, env, modelName);
    }
    return <div style={s.modalBackdrop} onClick={() => setModal(null)}>
            <div style={s.modalBox} onClick={e => e.stopPropagation()}>
              <div style={s.modalHeader}>
                <div style={s.modalTitle}>Benchmark commands</div>
                <button style={s.modalCloseBtn} onClick={() => setModal(null)} aria-label="Close">×</button>
              </div>
              <p style={{
      fontSize: 11,
      opacity: 0.7,
      margin: "0 0 12px"
    }}>
                For <code>{selSummary}</code>. Start the server with the Deploy command above, then run these against it.
              </p>

              {selAcc && <div>
                  <div style={s.sectionHeading}>Accuracy</div>
                  {bc.accuracy.length > 1 && <div style={s.benchChipRow}>
                      <span style={{
      fontSize: 11,
      opacity: 0.7
    }}>benchmark:</span>
                      {bc.accuracy.map(a => <button key={a.key} style={{
      ...s.benchChip,
      ...a.key === selAcc.key ? s.benchChipActive : {}
    }} onClick={() => setBenchAcc(a.key)}>
                          {a.label}
                        </button>)}
                    </div>}
                  <div style={{
      ...s.commandWrap,
      marginBottom: 6
    }}>
                    <div style={s.commandHeader}>
                      <div style={{
      fontSize: 11,
      opacity: 0.7
    }}>{selAcc.label}</div>
                      <button style={s.iconButton} onClick={() => copyBench("acc", accCmd)}>
                        {benchCopied === "acc" ? "✓ Copied" : "⧉ Copy"}
                      </button>
                    </div>
                    <pre style={s.commandPre}>{accCmd}</pre>
                  </div>
                  {bc.accuracy.length > 1 && <p style={{
      fontSize: 11,
      opacity: 0.7,
      margin: "0 0 4px"
    }}>
                      Switch the benchmark chip to see each eval's command.
                    </p>}
                </div>}

              {bc.speed && <div>
                  <div style={s.sectionHeading}>Speed</div>
                  {bc.speed.concurrencies.length > 1 && <div style={s.benchChipRow}>
                      <span style={{
      fontSize: 11,
      opacity: 0.7
    }}>max-concurrency:</span>
                      {bc.speed.concurrencies.map(c => <button key={c} style={{
      ...s.benchChip,
      ...c === selConc ? s.benchChipActive : {}
    }} onClick={() => setBenchConc(c)}>
                          {c}
                        </button>)}
                    </div>}
                  <div style={{
      ...s.commandWrap,
      marginBottom: 6
    }}>
                    <div style={s.commandHeader}>
                      <div style={{
      fontSize: 11,
      opacity: 0.7
    }}>max-concurrency = {selConc}</div>
                      <button style={s.iconButton} onClick={() => copyBench("speed", speedCmd)}>
                        {benchCopied === "speed" ? "✓ Copied" : "⧉ Copy"}
                      </button>
                    </div>
                    <pre style={s.commandPre}>{speedCmd}</pre>
                  </div>
                  <p style={{
      fontSize: 11,
      opacity: 0.7,
      margin: "0 0 4px"
    }}>
                    One command — switch the concurrency chip (or edit <code>--max-concurrency</code>) to reproduce each Speed column.
                  </p>
                </div>}

              <p style={{
      fontSize: 11,
      opacity: 0.7,
      marginTop: 12
    }}>
                Edit <code>CURL_HOST</code> / <code>CURL_PORT</code> in the Env panel.
              </p>
            </div>
          </div>;
  })()}
    </div>;
};

## Deployment

<a id="install" />

<Accordion title="Install SGLang">
  Use an SGLang build that includes GLM-5.3-Flash support.

  ```bash Command theme={null}
  docker pull lmsysorg/sglang:glm-5.3-flash
  ```

  The deployment panel can render a complete `docker run` command for the selected hardware and options. See [Install SGLang with Docker](/docs/get-started/install#method-3-using-docker) for host setup.
</Accordion>

Choose your hardware, then choose the operating point that matches your workload:

* **Low Latency** starts with MTP 5/1/6 speculative decoding and tensor parallelism to shorten interactive responses.
* **High Throughput** starts with speculative decoding off, which avoids draft-and-verify overhead under sustained batches.

Every listed hardware platform exposes both strategies. A **Verified** badge means that exact hardware and command were tested. **Final Verification In Progress** means the recipe runs and is queued for measurement on the final weights. **Not Verified** means the command is a supported starting point that still needs workload validation. A choice is disabled only when the underlying runtime combination is known to be unsupported.

The recommended selection is only a starting point. The same panel also lets you override the KV/DSA pairing, multimodal feature transport, Breakable Cuda Graph, and HiCache tiers. Changing an option that was not part of the measured command changes the badge to **Not Verified** without hiding the option.

**Breakable Cuda Graph** defaults to **Off**. Select **On** to add `--cuda-graph-backend-prefill breakable` to the generated command. This requires a build that includes [PR #38522](https://github.com/sgl-project/sglang/pull/38522).

<Deployment config={config} benchmarks={benchmarks} />

<Note>
  Generated commands leave `--mamba-full-memory-ratio` at its `0.9` default, which is a generic starting point rather than a workload-tuned split: too low starves the KDA state pool and clamps `max_running_requests`, too high over-provisions it and shrinks the KV pool. Use the repo-local [`compute-mamba-ratio`](https://github.com/sgl-project/sglang/blob/main/.claude/skills/compute-mamba-ratio/SKILL.md) skill to compute the balanced ratio — or the `--max-mamba-cache-size` pin to use instead — from your average request length and the two pool sizes printed in one boot log. See [Size both memory pools](#size-both-memory-pools) for what each pool caps.
</Note>

## Playground

Use the Playground for lower-level tuning such as attention parallelism, MoE communication, speculative decoding, and reasoning or tool parsers. It inherits every selection from the deployment panel and shows only the command-line diff.

<Playground config={config} />

## 1. Model introduction

GLM-5.3-Flash is a natively multimodal Mixture-of-Experts model built around a hybrid attention architecture — 320B total parameters with 18B active. Its 45 text layers combine MLA attention, DSA sparse attention, and KDA linear attention, and a 24-layer vision encoder handles image and video input. The checkpoint uses 288 routed experts with 8 active experts per token and includes a native MTP draft layer for speculative decoding. See the [GLM-5.3-Flash blog](https://z.ai/blog/glm-5.3-flash) and the [GLM-5 technical report](https://arxiv.org/abs/2602.15763) for training details.

<table style={{width: "100%", borderCollapse: "collapse", tableLayout: "fixed"}}>
  <colgroup>
    <col style={{width: "28%"}} />

    <col style={{width: "72%"}} />
  </colgroup>

  <thead>
    <tr style={{borderBottom: "2px solid #d55816"}}>
      <th style={{textAlign: "left", padding: "10px 12px", fontWeight: 700}}>Attribute</th>
      <th style={{textAlign: "left", padding: "10px 12px", fontWeight: 700}}>Description</th>
    </tr>
  </thead>

  <tbody>
    <tr>
      <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Architecture</td>
      <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>MoE with hybrid attention (MLA, DSA, KDA), mHC, and MTP — 320B total / 18B active parameters</td>
    </tr>

    <tr>
      <td style={{padding: "9px 12px"}}>Precision</td>
      <td style={{padding: "9px 12px"}}>FP8 weights; FP8 KV cache by default on Blackwell, BF16 KV cache on H100 and H200</td>
    </tr>

    <tr>
      <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Context</td>
      <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>1M tokens</td>
    </tr>

    <tr>
      <td style={{padding: "9px 12px"}}>Inputs</td>
      <td style={{padding: "9px 12px"}}>Text, image, and video</td>
    </tr>

    <tr>
      <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Generation defaults</td>
      <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}><code>temperature=1.0</code>, <code>top\_p=0.95</code>, thinking enabled</td>
    </tr>
  </tbody>
</table>

The deployment recipes use the checkpoint's generation configuration. Override sampling only when your application has its own evaluated settings.

## 2. Configuration tips

### Choose a strategy

Start with **Low Latency** for chat and agent workloads. It drafts from the checkpoint's MTP head at a fixed depth (5 steps, top-k 1, 6 draft tokens) with natural acceptance. Measure **High Throughput** for heavily batched traffic where disabling speculative decoding can be more efficient. SGLang serves MTP through `--speculative-algorithm EAGLE` (upstream folds the older NEXTN spelling into EAGLE), so generated commands use that flag value.

Strategy labels describe the workload goal, not a hardware restriction. Both strategies stay available when you switch hardware; only the verification badge changes.

### Change the speculative algorithm

The **Speculative** card in the Playground changes the algorithm without leaving the selected strategy:

* **EAGLE / MTP 5-1-6** is exactly what Low Latency serves, so a Low Latency base starts on this chip. Pick it from a High Throughput base to keep that recipe's other settings and add the MTP head.
* **Off (greedy)** strips the whole `--speculative-*` family, which is what High Throughput already starts from.
* **DFlash2** swaps the in-checkpoint MTP head for the trained block-diffusion draft in [`incoai/GLM-5.3-Flash-DFlash2`](https://huggingface.co/incoai/GLM-5.3-Flash-DFlash2). The draft proposes a whole block per step and the target verifies it in one forward pass, so output quality stays the target's. Its block size comes from the draft checkpoint, and the draft runs on `fa4` rather than the target's DSA backends. It needs a build that carries the GLM-5.3-Flash hidden-state capture from [PR #36708](https://github.com/sgl-project/sglang/pull/36708), which is merged into the [PR #36507](https://github.com/sgl-project/sglang/pull/36507) support branch (`xinyuan/glm-5.3-flash-support`) rather than into `main`, so the image pinned above is not enough on its own — pull that branch at its current head, or add #36708's commit on top of an older checkout. The draft repository is also access-gated: request access on its model page, then download it alongside the target before serving. This combination is not yet measured on the cookbook hardware, so treat it as a starting point.

Neither algorithm runs with DP-Attention; the card disables the affected chips and names the reason.

### Size both memory pools

GLM-5.3-Flash maintains a paged KV pool for attention and a separate KDA state pool. The KDA state pool can limit concurrency before the KV pool is full. If startup reduces `max_running_requests` because of KDA state capacity, increase `--mamba-full-memory-ratio` or set `--max-mamba-cache-size` for the expected concurrency, then tune `--max-running-requests` to the workload.

Keep the prefix cache enabled for every strategy.

Keep the checkpoint's KDA lower-bound setting unchanged. In particular, do not override `linear_lower_bound` through `--json-model-override-args`.

### Keep the KV and DSA backends paired

On Blackwell, the recipes default to an FP8 KV cache with TRT-LLM DSA: on GB300 this pairing measured 2.9–5.7% higher throughput and about 1.8x the KV token capacity at identical pool bytes, with GSM8K accuracy within noise of BF16. BF16 KV with TileLang DSA remains selectable in the deployment panel and is the default on H100 and H200, where FP8 KV with TRT-LLM DSA is disabled. Switch the dtype and both DSA backends together; TileLang DSA with FP8 KV is not a valid CUDA combination.

### Decode context parallelism

Decode context parallelism (DCP) shards the KV across GPUs during decode to cut long-context latency. The **Context Parallelism** row offers DCP4, validated on 4x GB300 with both KV pairings on the earlier TP4/EP4 recipe and queued for re-measurement on the current command; it requires the current release image, which carries the TileLang LSE fix. Other platforms and attention backends are unvalidated, and draft-extend v2 is unsupported under DCP.

### Extend the cache hierarchy

Keep **HiCache** off when GPU memory is sufficient. Select **L1 + L2** to spill reusable cache entries into host memory. Select **+ L3** only after configuring Mooncake on every serving node; the generated command exposes the required configuration path. These options remain selectable but are marked **Not Verified** until the resulting command is validated on the chosen hardware.

### Multimodal memory

All strategies enable multimodal serving. The processor samples video at 2 FPS and caps video input at 240,000 visual tokens. Install `torchcodec` in the serving environment before sending video requests. For very long videos on 4x GB300, use encoder disaggregation to isolate the vision encoder's memory spikes from language decoding.

The default multimodal feature transport is automatic, and on a single CUDA node auto resolves to CPU transport. CUDA IPC is opt-in: pass `--mm-feature-transport cuda_ipc` when lower transfer latency matters more than the GPU memory the IPC pool reserves. CUDA VMM transport applies only to multi-node GB200/GB300 systems on the MNNVL fabric, where auto selects it.

## 3. Advanced usage

### 3.1 Reasoning

Thinking is enabled by the checkpoint's generation configuration, and generated commands enable `--reasoning-parser glm45` by default. The OpenAI-compatible API then places thinking in `message.reasoning_content` and the final answer in `message.content`. You can disable **Reasoning Parser** in the Playground when an integration needs the raw response format.

To disable thinking for a request, pass `chat_template_kwargs: {"thinking": false}` in the request body.

### 3.2 Tool calling

Generated commands enable `--tool-call-parser glm47` by default, so structured calls are returned in `message.tool_calls`. You can disable **Tool Call Parser** in the Playground when tool calling is not needed. On follow-up turns, read both `reasoning_content` and `content` because a thinking model can use either field around tool execution.

### 3.3 Multimodal serving

The base recipes accept image and video content through the OpenAI-compatible chat API. Keep the processor defaults unless you have measured a different sampling or resize policy. Inputs above the video token budget are clamped to the processor's limit rather than rejected.

### 3.4 Encoder disaggregation

Encoder disaggregation separates vision preprocessing from language inference. The verified topology uses one 4x GB300 node shared by an encoder-only TP4 process on port 30001 and a language-only TP4 process on port 30000. Start the encoder first.

<Accordion title="Encoder server (GB300)">
  ```bash Command theme={null}
  sglang serve \
    --model-path zai-org/GLM-5.3-Flash \
    --tp-size 4 \
    --encoder-only \
    --host 0.0.0.0 \
    --port 30001
  ```
</Accordion>

<Accordion title="Language server (GB300)">
  ```bash Command theme={null}
  sglang serve \
    --model-path zai-org/GLM-5.3-Flash \
    --tp-size 4 \
    --attention-backend dsa \
    --dsa-prefill-backend tilelang \
    --dsa-decode-backend tilelang \
    --linear-attn-backend triton \
    --kv-cache-dtype bfloat16 \
    --quantization fp8 \
    --moe-runner-backend flashinfer_trtllm \
    --max-running-requests 64 \
    --chunked-prefill-size 8192 \
    --max-prefill-tokens 8192 \
    --disable-prefill-cuda-graph \
    --speculative-algorithm EAGLE \
    --speculative-num-steps 3 \
    --speculative-eagle-topk 1 \
    --speculative-num-draft-tokens 4 \
    --language-only \
    --encoder-urls http://localhost:30001 \
    --mem-fraction-static 0.78 \
    --host 0.0.0.0 \
    --port 30000
  ```
</Accordion>

This topology served image requests and videos up to 238,080 visual tokens. In a concurrent long-video workload, the largest observed decode gap fell from 5.53 seconds in unified serving to 1.79 seconds with encoder disaggregation. Keep `--mem-fraction-static 0.78` on the language process so the encoder retains room for vision workspaces.

Install `torchcodec` for video and see the [encoder disaggregation guide](/docs/advanced_features/epd_disaggregation) for the generic architecture and operational model.

### 3.5 PD disaggregation (preview)

PD splits prefill and decode into separate server groups behind a router. For this hybrid model, the transfer moves both the paged DSA KV and the KDA recurrent state.

<Warning>
  PD serving is mechanically validated with dummy weights only — startup, bootstrap, state transfer, and request flow all work on 4x GB300. It has not been load- or accuracy-tested. Treat it as a preview until the real-weight gate completes.
</Warning>

<Accordion title="PD serving on 4x GB300 (single node)">
  ```bash Prefill (GPU 0-1) theme={null}
  sglang serve \
    --model-path zai-org/GLM-5.3-Flash \
    --tp-size 2 \
    --dsa-prefill-backend tilelang \
    --dsa-decode-backend tilelang \
    --kv-cache-dtype bfloat16 \
    --moe-runner-backend triton \
    --disaggregation-mode prefill \
    --disaggregation-bootstrap-port 8998 \
    --disaggregation-transfer-backend nixl \
    --host 0.0.0.0 \
    --port 31000
  ```

  ```bash Decode (GPU 2-3) theme={null}
  sglang serve \
    --model-path zai-org/GLM-5.3-Flash \
    --tp-size 2 \
    --base-gpu-id 2 \
    --dsa-prefill-backend tilelang \
    --dsa-decode-backend tilelang \
    --kv-cache-dtype bfloat16 \
    --moe-runner-backend triton \
    --disaggregation-mode decode \
    --disaggregation-transfer-backend nixl \
    --host 0.0.0.0 \
    --port 32000
  ```

  ```bash Router theme={null}
  python -m sglang_router.launch_router \
    --pd-disaggregation \
    --mini-lb \
    --prefill http://127.0.0.1:31000 8998 \
    --decode http://127.0.0.1:32000 \
    --host 0.0.0.0 \
    --port 30000
  ```

  The positional `8998` after `--prefill` must equal the prefill server's `--disaggregation-bootstrap-port`.
</Accordion>

Operational notes:

* Give each role a distinct `--nccl-port` when both share one node.
* Single-node NIXL needs `UCX_NET_DEVICES=lo` and `UCX_TLS=tcp,cuda_copy,cuda_ipc,self,sm` in both server environments.
* The validated arm used the triton MoE runner; the flashinfer\_trtllm runner from the deployment recipes is untested under PD.

Known limitations:

* Speculative decoding does not start under PD at the current cut (draft-graph capture width assert on the prefill role; decode-side memory pressure at TP2). Run PD without speculative flags.
* Prefill and decode with different TP sizes transfer state through slice paths, but numeric correctness is unverified. Keep both roles at the same TP size.
