{
  "schemaVersion": 1,
  "runs": [
    {
      "id": "2026-10-21-m5max-ornith-observed-subset",
      "provenance": {
        "startedAt": "2026-10-21T09:53:34.651Z",
        "device": {
          "label": "Mac · Apple M5 Max",
          "platform": "darwin",
          "arch": "arm64",
          "memoryGb": 128,
          "osRelease": "darwin 27.0.0",
          "cpuModel": "Apple M5 Max",
          "gpuModel": "Apple M5 Max (integrated GPU)"
        },
        "harnessCommit": "bc4af5dc2649",
        "sourceHash": "d68acf8def595c54deea3468ab9efe7ff2c6970be4e1d64e6fe034dbb82dd042",
        "serviceBuildHash": "3acca0c382dfe21a3194e9e27a77e51b47e5cfb46e1ecd5af8a0b000f6754241",
        "catalogHash": "a95cdb382c0974f495a3f868a17b0b450a27ffa7dbfe472f6b82c668bcde1d4a",
        "gildeVersion": "0.2.3",
        "count": 3,
        "judgeModelId": null,
        "engineBuilds": {
          "llama-cpp": "d81235049384534c167caea52b85a694f6103d14"
        }
      },
      "scope": "subset",
      "memoryCoverage": "full",
      "suites": [
        "core",
        "productivity",
        "developer",
        "complex-work"
      ],
      "scenariosBySuite": {
        "core": [
          "tictactoe",
          "petshop",
          "tankcombat",
          "schema-migration",
          "failing-tests-spec",
          "symptom-debug",
          "data-wrangle",
          "incident-postmortem",
          "ops-runbook-anomaly",
          "plan-and-estimate",
          "conflict-synthesis"
        ],
        "productivity": [
          "constrained-comms",
          "records-intake",
          "plan-and-estimate",
          "meeting-followup",
          "craftbook-spreadsheet-model",
          "conflict-synthesis",
          "docblocks-theme-roundtrip",
          "craftbook-powerpoint-deck",
          "wikipedia-research-brief",
          "craftbook-week-plan",
          "craftbook-ab-test-readout",
          "craftbook-annotated-bibliography",
          "craftbook-research-to-document"
        ],
        "developer": [
          "fix-squisq-bugs",
          "dev-craftbook-routing",
          "interface-contract",
          "craftbook-code-review",
          "craftbook-api-contract-review",
          "craftbook-deep-security-review",
          "large-pr-review",
          "craftbook-codemod-sweep",
          "craftbook-refactor-module",
          "craftbook-bug-fix-tdd"
        ],
        "complex-work": [
          "craftbook-find-vs-create",
          "craftbook-edit-midtask",
          "craftbook-invoice-run",
          "craftbook-author-linear",
          "craftbook-route-multi",
          "craftbook-export-generalize",
          "craftbook-author-params",
          "craftbook-author-gate-script",
          "craftbook-author-fanout"
        ]
      },
      "note": "PARTIAL OBSERVATIONS: Ornith recorded 85 of 129 planned trials; 84 stable-identity observations are included here. The final drift-flagged failure is retained in the 601-observation companion and omitted from these comparison cells. No third repetition was run; report counts, not rates. Nemotron and MiniCPM were not run. Four suites, 43 suite entries, three requested repetitions. Fixed classifications and scores are unchanged. No pooled qualitative judge score: see scorecard-details.json for exact per-task rubrics and 248 artifact reviews. Performance columns are arithmetic means of short admission probes (each under 256 decode tokens), not sustained in-trial speeds; prefer the companion in-trial distributions for throughput comparisons. Six frozen infrastructure exclusions remain recorded across the package. Models ran sequentially on one device, with model-specific frozen tuning."
    },
    {
      "id": "2026-10-16-m5max-local-seven-completed-arms",
      "provenance": {
        "startedAt": "2026-10-16T04:34:42.565Z",
        "device": {
          "label": "Mac · Apple M5 Max",
          "platform": "darwin",
          "arch": "arm64",
          "memoryGb": 128,
          "osRelease": "darwin 27.0.0",
          "cpuModel": "Apple M5 Max",
          "gpuModel": "Apple M5 Max (integrated GPU)"
        },
        "harnessCommit": "bc4af5dc2649",
        "sourceHash": "d68acf8def595c54deea3468ab9efe7ff2c6970be4e1d64e6fe034dbb82dd042",
        "serviceBuildHash": "3acca0c382dfe21a3194e9e27a77e51b47e5cfb46e1ecd5af8a0b000f6754241",
        "catalogHash": "a95cdb382c0974f495a3f868a17b0b450a27ffa7dbfe472f6b82c668bcde1d4a",
        "gildeVersion": "0.2.3",
        "count": 3,
        "judgeModelId": null,
        "engineBuilds": {
          "llama-cpp": "d81235049384534c167caea52b85a694f6103d14"
        }
      },
      "scope": "suite",
      "memoryCoverage": "full",
      "suites": [
        "core",
        "productivity",
        "developer",
        "complex-work"
      ],
      "scenariosBySuite": {
        "core": [
          "tictactoe",
          "petshop",
          "tankcombat",
          "schema-migration",
          "failing-tests-spec",
          "symptom-debug",
          "data-wrangle",
          "incident-postmortem",
          "ops-runbook-anomaly",
          "plan-and-estimate",
          "conflict-synthesis"
        ],
        "productivity": [
          "constrained-comms",
          "records-intake",
          "plan-and-estimate",
          "meeting-followup",
          "craftbook-spreadsheet-model",
          "conflict-synthesis",
          "docblocks-theme-roundtrip",
          "craftbook-powerpoint-deck",
          "wikipedia-research-brief",
          "craftbook-week-plan",
          "craftbook-ab-test-readout",
          "craftbook-annotated-bibliography",
          "craftbook-research-to-document"
        ],
        "developer": [
          "fix-squisq-bugs",
          "dev-craftbook-routing",
          "interface-contract",
          "craftbook-code-review",
          "craftbook-api-contract-review",
          "craftbook-deep-security-review",
          "large-pr-review",
          "craftbook-codemod-sweep",
          "craftbook-refactor-module",
          "craftbook-bug-fix-tdd"
        ],
        "complex-work": [
          "craftbook-find-vs-create",
          "craftbook-edit-midtask",
          "craftbook-invoice-run",
          "craftbook-author-linear",
          "craftbook-route-multi",
          "craftbook-export-generalize",
          "craftbook-author-params",
          "craftbook-author-gate-script",
          "craftbook-author-fanout"
        ]
      },
      "note": "Completed arms only: LFM 2.5, Qwen 3.6, Qwen 3.8 and BTL4 each recorded all 129 trials (516 observations). Ornith is retained in a separate partial round, not mixed into this headline table. Four suites, 43 suite entries, three requested repetitions. Fixed classifications and scores are unchanged. No pooled qualitative judge score: see scorecard-details.json for exact per-task rubrics and 248 artifact reviews. Performance columns are arithmetic means of short admission probes (each under 256 decode tokens), not sustained in-trial speeds; prefer the companion in-trial distributions for throughput comparisons. Six frozen infrastructure exclusions remain recorded across the package. Models ran sequentially on one device, with model-specific frozen tuning."
    }
  ],
  "results": [
    {
      "modelId": "ornith1.5-35b-a3b-q4",
      "label": "Ornith 1.5 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-21-m5max-ornith-observed-subset",
      "suiteId": "complex-work",
      "performance": {
        "prefillTokensPerSec": 2619,
        "decodeTokensPerSec": 190.6,
        "samples": 1
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 37090,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "craftbook-find-vs-create",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 264559
        },
        {
          "scenarioId": "craftbook-edit-midtask",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 837923
        },
        {
          "scenarioId": "craftbook-invoice-run",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 457599
        },
        {
          "scenarioId": "craftbook-author-linear",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 2313716
        },
        {
          "scenarioId": "craftbook-route-multi",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 408666
        },
        {
          "scenarioId": "craftbook-export-generalize",
          "trials": 2,
          "successes": 0,
          "nonModelFailures": 1,
          "medianDurationMs": 1049764
        },
        {
          "scenarioId": "craftbook-author-params",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 4396658
        },
        {
          "scenarioId": "craftbook-author-gate-script",
          "trials": 1,
          "successes": 0,
          "nonModelFailures": 1,
          "medianDurationMs": 3849165
        },
        {
          "scenarioId": "craftbook-author-fanout",
          "trials": 1,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 3250134
        }
      ]
    },
    {
      "modelId": "ornith1.5-35b-a3b-q4",
      "label": "Ornith 1.5 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-21-m5max-ornith-observed-subset",
      "suiteId": "core",
      "performance": {
        "prefillTokensPerSec": 2619,
        "decodeTokensPerSec": 190.6,
        "samples": 1
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 29403,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "tictactoe",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 273114
        },
        {
          "scenarioId": "petshop",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 343125
        },
        {
          "scenarioId": "tankcombat",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 287672
        },
        {
          "scenarioId": "schema-migration",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 331817
        },
        {
          "scenarioId": "failing-tests-spec",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 412958
        },
        {
          "scenarioId": "symptom-debug",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 282311
        },
        {
          "scenarioId": "data-wrangle",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 287391
        },
        {
          "scenarioId": "incident-postmortem",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 323081
        },
        {
          "scenarioId": "ops-runbook-anomaly",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 272756
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 262619
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 2,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 330480
        }
      ]
    },
    {
      "modelId": "ornith1.5-35b-a3b-q4",
      "label": "Ornith 1.5 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-21-m5max-ornith-observed-subset",
      "suiteId": "developer",
      "performance": {
        "prefillTokensPerSec": 2619,
        "decodeTokensPerSec": 190.6,
        "samples": 1
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 33246,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "fix-squisq-bugs",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 295664
        },
        {
          "scenarioId": "dev-craftbook-routing",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 297156
        },
        {
          "scenarioId": "interface-contract",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 489689
        },
        {
          "scenarioId": "craftbook-code-review",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 414169
        },
        {
          "scenarioId": "craftbook-api-contract-review",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 645975
        },
        {
          "scenarioId": "craftbook-deep-security-review",
          "trials": 2,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 393937
        },
        {
          "scenarioId": "large-pr-review",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 845216
        },
        {
          "scenarioId": "craftbook-codemod-sweep",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 415800
        },
        {
          "scenarioId": "craftbook-refactor-module",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 1133039
        },
        {
          "scenarioId": "craftbook-bug-fix-tdd",
          "trials": 2,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1100134
        }
      ]
    },
    {
      "modelId": "ornith1.5-35b-a3b-q4",
      "label": "Ornith 1.5 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-21-m5max-ornith-observed-subset",
      "suiteId": "productivity",
      "performance": {
        "prefillTokensPerSec": 2619,
        "decodeTokensPerSec": 190.6,
        "samples": 1
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 33083,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "constrained-comms",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 259926
        },
        {
          "scenarioId": "records-intake",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 521559
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 265230
        },
        {
          "scenarioId": "meeting-followup",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 353050
        },
        {
          "scenarioId": "craftbook-spreadsheet-model",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 1115483
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 2,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 422319
        },
        {
          "scenarioId": "docblocks-theme-roundtrip",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 282786
        },
        {
          "scenarioId": "craftbook-powerpoint-deck",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 431923
        },
        {
          "scenarioId": "wikipedia-research-brief",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 345492
        },
        {
          "scenarioId": "craftbook-week-plan",
          "trials": 2,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 3324322
        },
        {
          "scenarioId": "craftbook-ab-test-readout",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 843111
        },
        {
          "scenarioId": "craftbook-annotated-bibliography",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 467505
        },
        {
          "scenarioId": "craftbook-research-to-document",
          "trials": 2,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 360697
        }
      ]
    },
    {
      "modelId": "btl4-35b-q4",
      "label": "BTL-4 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "complex-work",
      "performance": {
        "prefillTokensPerSec": 2828,
        "decodeTokensPerSec": 119.4,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 34327,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "craftbook-find-vs-create",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 258512
        },
        {
          "scenarioId": "craftbook-edit-midtask",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 3024990
        },
        {
          "scenarioId": "craftbook-invoice-run",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 516217
        },
        {
          "scenarioId": "craftbook-author-linear",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 1571348
        },
        {
          "scenarioId": "craftbook-route-multi",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 911631
        },
        {
          "scenarioId": "craftbook-export-generalize",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1595515
        },
        {
          "scenarioId": "craftbook-author-params",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1859838
        },
        {
          "scenarioId": "craftbook-author-gate-script",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 1,
          "medianDurationMs": 1933159
        },
        {
          "scenarioId": "craftbook-author-fanout",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 3133753
        }
      ]
    },
    {
      "modelId": "lfm2.5-24b-a2b-q4",
      "label": "LFM2.5 (24B-A2B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "24B-A2B",
      "quantization": "Q4_0",
      "grantedContextTokens": 131072,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "complex-work",
      "performance": {
        "prefillTokensPerSec": 4711,
        "decodeTokensPerSec": 182.6,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 131072,
        "peakMemoryMb": 18257,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "craftbook-find-vs-create",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 31765
        },
        {
          "scenarioId": "craftbook-edit-midtask",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 1085520
        },
        {
          "scenarioId": "craftbook-invoice-run",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 363799
        },
        {
          "scenarioId": "craftbook-author-linear",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 832043
        },
        {
          "scenarioId": "craftbook-route-multi",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1265021
        },
        {
          "scenarioId": "craftbook-export-generalize",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 1143070
        },
        {
          "scenarioId": "craftbook-author-params",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 1596785
        },
        {
          "scenarioId": "craftbook-author-gate-script",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 2196696
        },
        {
          "scenarioId": "craftbook-author-fanout",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 832587
        }
      ]
    },
    {
      "modelId": "qwen3.6-35b-a3b-q4",
      "label": "Qwen 3.6 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "UD-Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "complex-work",
      "performance": {
        "prefillTokensPerSec": 1441,
        "decodeTokensPerSec": 133.9,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 38879,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "craftbook-find-vs-create",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 31792
        },
        {
          "scenarioId": "craftbook-edit-midtask",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 524110
        },
        {
          "scenarioId": "craftbook-invoice-run",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 252112
        },
        {
          "scenarioId": "craftbook-author-linear",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 1224440
        },
        {
          "scenarioId": "craftbook-route-multi",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 354616
        },
        {
          "scenarioId": "craftbook-export-generalize",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 1541391
        },
        {
          "scenarioId": "craftbook-author-params",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1416483
        },
        {
          "scenarioId": "craftbook-author-gate-script",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 1937914
        },
        {
          "scenarioId": "craftbook-author-fanout",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 2,
          "medianDurationMs": 1622412
        }
      ]
    },
    {
      "modelId": "qwen3.8-27b-q4",
      "label": "Qwen 3.8 (27B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "27B",
      "quantization": "UD-Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "complex-work",
      "performance": {
        "prefillTokensPerSec": 501,
        "decodeTokensPerSec": 36.7,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 48108,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "craftbook-find-vs-create",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 71027
        },
        {
          "scenarioId": "craftbook-edit-midtask",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 1660658
        },
        {
          "scenarioId": "craftbook-invoice-run",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 641101
        },
        {
          "scenarioId": "craftbook-author-linear",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 6006562
        },
        {
          "scenarioId": "craftbook-route-multi",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 1973780
        },
        {
          "scenarioId": "craftbook-export-generalize",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 2560265
        },
        {
          "scenarioId": "craftbook-author-params",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 2802336
        },
        {
          "scenarioId": "craftbook-author-gate-script",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 1297159
        },
        {
          "scenarioId": "craftbook-author-fanout",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 3424360
        }
      ]
    },
    {
      "modelId": "btl4-35b-q4",
      "label": "BTL-4 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "core",
      "performance": {
        "prefillTokensPerSec": 2828,
        "decodeTokensPerSec": 119.4,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 28769,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "tictactoe",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 268769
        },
        {
          "scenarioId": "petshop",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 329531
        },
        {
          "scenarioId": "tankcombat",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 329231
        },
        {
          "scenarioId": "schema-migration",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 311351
        },
        {
          "scenarioId": "failing-tests-spec",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 283708
        },
        {
          "scenarioId": "symptom-debug",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 264349
        },
        {
          "scenarioId": "data-wrangle",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 281179
        },
        {
          "scenarioId": "incident-postmortem",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 438478
        },
        {
          "scenarioId": "ops-runbook-anomaly",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 286291
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 274756
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 381289
        }
      ]
    },
    {
      "modelId": "lfm2.5-24b-a2b-q4",
      "label": "LFM2.5 (24B-A2B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "24B-A2B",
      "quantization": "Q4_0",
      "grantedContextTokens": 131072,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "core",
      "performance": {
        "prefillTokensPerSec": 4711,
        "decodeTokensPerSec": 182.6,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 131072,
        "peakMemoryMb": 18095,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "tictactoe",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 116640
        },
        {
          "scenarioId": "petshop",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 109150
        },
        {
          "scenarioId": "tankcombat",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 517713
        },
        {
          "scenarioId": "schema-migration",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 186117
        },
        {
          "scenarioId": "failing-tests-spec",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 581277
        },
        {
          "scenarioId": "symptom-debug",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 67285
        },
        {
          "scenarioId": "data-wrangle",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 1,
          "medianDurationMs": 956717
        },
        {
          "scenarioId": "incident-postmortem",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 273746
        },
        {
          "scenarioId": "ops-runbook-anomaly",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 37456
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 57634
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 395032
        }
      ]
    },
    {
      "modelId": "qwen3.6-35b-a3b-q4",
      "label": "Qwen 3.6 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "UD-Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "core",
      "performance": {
        "prefillTokensPerSec": 1441,
        "decodeTokensPerSec": 133.9,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 30277,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "tictactoe",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 39393
        },
        {
          "scenarioId": "petshop",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 64270
        },
        {
          "scenarioId": "tankcombat",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 49993
        },
        {
          "scenarioId": "schema-migration",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 79081
        },
        {
          "scenarioId": "failing-tests-spec",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 22849
        },
        {
          "scenarioId": "symptom-debug",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 37125
        },
        {
          "scenarioId": "data-wrangle",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 57897
        },
        {
          "scenarioId": "incident-postmortem",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 217947
        },
        {
          "scenarioId": "ops-runbook-anomaly",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 57165
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 32053
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 162855
        }
      ]
    },
    {
      "modelId": "qwen3.8-27b-q4",
      "label": "Qwen 3.8 (27B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "27B",
      "quantization": "UD-Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "core",
      "performance": {
        "prefillTokensPerSec": 501,
        "decodeTokensPerSec": 36.7,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 41395,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "tictactoe",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 150397
        },
        {
          "scenarioId": "petshop",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 230265
        },
        {
          "scenarioId": "tankcombat",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 271513
        },
        {
          "scenarioId": "schema-migration",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 254058
        },
        {
          "scenarioId": "failing-tests-spec",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 53096
        },
        {
          "scenarioId": "symptom-debug",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 62531
        },
        {
          "scenarioId": "data-wrangle",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 97967
        },
        {
          "scenarioId": "incident-postmortem",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 907131
        },
        {
          "scenarioId": "ops-runbook-anomaly",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 127767
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 142843
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 404016
        }
      ]
    },
    {
      "modelId": "btl4-35b-q4",
      "label": "BTL-4 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "developer",
      "performance": {
        "prefillTokensPerSec": 2828,
        "decodeTokensPerSec": 119.4,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 29953,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "fix-squisq-bugs",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 276241
        },
        {
          "scenarioId": "dev-craftbook-routing",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 255739
        },
        {
          "scenarioId": "interface-contract",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 265604
        },
        {
          "scenarioId": "craftbook-code-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 306338
        },
        {
          "scenarioId": "craftbook-api-contract-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 723513
        },
        {
          "scenarioId": "craftbook-deep-security-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 317065
        },
        {
          "scenarioId": "large-pr-review",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1217188
        },
        {
          "scenarioId": "craftbook-codemod-sweep",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 578769
        },
        {
          "scenarioId": "craftbook-refactor-module",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1784954
        },
        {
          "scenarioId": "craftbook-bug-fix-tdd",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 462914
        }
      ]
    },
    {
      "modelId": "lfm2.5-24b-a2b-q4",
      "label": "LFM2.5 (24B-A2B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "24B-A2B",
      "quantization": "Q4_0",
      "grantedContextTokens": 131072,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "developer",
      "performance": {
        "prefillTokensPerSec": 4711,
        "decodeTokensPerSec": 182.6,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 131072,
        "peakMemoryMb": 19568,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "fix-squisq-bugs",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 752529
        },
        {
          "scenarioId": "dev-craftbook-routing",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 38163
        },
        {
          "scenarioId": "interface-contract",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 310880
        },
        {
          "scenarioId": "craftbook-code-review",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 629491
        },
        {
          "scenarioId": "craftbook-api-contract-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 693116
        },
        {
          "scenarioId": "craftbook-deep-security-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 313233
        },
        {
          "scenarioId": "large-pr-review",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 996632
        },
        {
          "scenarioId": "craftbook-codemod-sweep",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 556275
        },
        {
          "scenarioId": "craftbook-refactor-module",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1335708
        },
        {
          "scenarioId": "craftbook-bug-fix-tdd",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 1029543
        }
      ]
    },
    {
      "modelId": "qwen3.6-35b-a3b-q4",
      "label": "Qwen 3.6 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "UD-Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "developer",
      "performance": {
        "prefillTokensPerSec": 1441,
        "decodeTokensPerSec": 133.9,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 34771,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "fix-squisq-bugs",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 42753
        },
        {
          "scenarioId": "dev-craftbook-routing",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 27412
        },
        {
          "scenarioId": "interface-contract",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 42054
        },
        {
          "scenarioId": "craftbook-code-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 142568
        },
        {
          "scenarioId": "craftbook-api-contract-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 408920
        },
        {
          "scenarioId": "craftbook-deep-security-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 128784
        },
        {
          "scenarioId": "large-pr-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 636035
        },
        {
          "scenarioId": "craftbook-codemod-sweep",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 281156
        },
        {
          "scenarioId": "craftbook-refactor-module",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 1495007
        },
        {
          "scenarioId": "craftbook-bug-fix-tdd",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 312367
        }
      ]
    },
    {
      "modelId": "qwen3.8-27b-q4",
      "label": "Qwen 3.8 (27B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "27B",
      "quantization": "UD-Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "developer",
      "performance": {
        "prefillTokensPerSec": 501,
        "decodeTokensPerSec": 36.7,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 46496,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "fix-squisq-bugs",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 140022
        },
        {
          "scenarioId": "dev-craftbook-routing",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 78158
        },
        {
          "scenarioId": "interface-contract",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 280371
        },
        {
          "scenarioId": "craftbook-code-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 345949
        },
        {
          "scenarioId": "craftbook-api-contract-review",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 1906427
        },
        {
          "scenarioId": "craftbook-deep-security-review",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 366696
        },
        {
          "scenarioId": "large-pr-review",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 2624327
        },
        {
          "scenarioId": "craftbook-codemod-sweep",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 823611
        },
        {
          "scenarioId": "craftbook-refactor-module",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 2174900
        },
        {
          "scenarioId": "craftbook-bug-fix-tdd",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 592677
        }
      ]
    },
    {
      "modelId": "btl4-35b-q4",
      "label": "BTL-4 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "productivity",
      "performance": {
        "prefillTokensPerSec": 2828,
        "decodeTokensPerSec": 119.4,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 30710,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "constrained-comms",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 267131
        },
        {
          "scenarioId": "records-intake",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 376543
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 260999
        },
        {
          "scenarioId": "meeting-followup",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 346846
        },
        {
          "scenarioId": "craftbook-spreadsheet-model",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 461222
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 362254
        },
        {
          "scenarioId": "docblocks-theme-roundtrip",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 261586
        },
        {
          "scenarioId": "craftbook-powerpoint-deck",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 406963
        },
        {
          "scenarioId": "wikipedia-research-brief",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 510548
        },
        {
          "scenarioId": "craftbook-week-plan",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 5638451
        },
        {
          "scenarioId": "craftbook-ab-test-readout",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 627623
        },
        {
          "scenarioId": "craftbook-annotated-bibliography",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 430534
        },
        {
          "scenarioId": "craftbook-research-to-document",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 366807
        }
      ]
    },
    {
      "modelId": "lfm2.5-24b-a2b-q4",
      "label": "LFM2.5 (24B-A2B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "24B-A2B",
      "quantization": "Q4_0",
      "grantedContextTokens": 131072,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "productivity",
      "performance": {
        "prefillTokensPerSec": 4711,
        "decodeTokensPerSec": 182.6,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 131072,
        "peakMemoryMb": 18520,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "constrained-comms",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 68512
        },
        {
          "scenarioId": "records-intake",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 171320
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 88805
        },
        {
          "scenarioId": "meeting-followup",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 144722
        },
        {
          "scenarioId": "craftbook-spreadsheet-model",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 764317
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 327234
        },
        {
          "scenarioId": "docblocks-theme-roundtrip",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 94208
        },
        {
          "scenarioId": "craftbook-powerpoint-deck",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 719674
        },
        {
          "scenarioId": "wikipedia-research-brief",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 372891
        },
        {
          "scenarioId": "craftbook-week-plan",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 4459675
        },
        {
          "scenarioId": "craftbook-ab-test-readout",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 679074
        },
        {
          "scenarioId": "craftbook-annotated-bibliography",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 240960
        },
        {
          "scenarioId": "craftbook-research-to-document",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 259978
        }
      ]
    },
    {
      "modelId": "qwen3.6-35b-a3b-q4",
      "label": "Qwen 3.6 (35B-A3B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "35B-A3B",
      "quantization": "UD-Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "productivity",
      "performance": {
        "prefillTokensPerSec": 1441,
        "decodeTokensPerSec": 133.9,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 33475,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "constrained-comms",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 42128
        },
        {
          "scenarioId": "records-intake",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 87344
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 32070
        },
        {
          "scenarioId": "meeting-followup",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 118217
        },
        {
          "scenarioId": "craftbook-spreadsheet-model",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 204768
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 218254
        },
        {
          "scenarioId": "docblocks-theme-roundtrip",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 37138
        },
        {
          "scenarioId": "craftbook-powerpoint-deck",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 157928
        },
        {
          "scenarioId": "wikipedia-research-brief",
          "trials": 3,
          "successes": 0,
          "nonModelFailures": 0,
          "medianDurationMs": 298638
        },
        {
          "scenarioId": "craftbook-week-plan",
          "trials": 3,
          "successes": 2,
          "nonModelFailures": 0,
          "medianDurationMs": 3108960
        },
        {
          "scenarioId": "craftbook-ab-test-readout",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 315340
        },
        {
          "scenarioId": "craftbook-annotated-bibliography",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 162747
        },
        {
          "scenarioId": "craftbook-research-to-document",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 143413
        }
      ]
    },
    {
      "modelId": "qwen3.8-27b-q4",
      "label": "Qwen 3.8 (27B, Q4)",
      "engine": "llama-cpp",
      "tier": "medium",
      "parameterSize": "27B",
      "quantization": "UD-Q4_K_M",
      "grantedContextTokens": 262144,
      "runId": "2026-10-16-m5max-local-seven-completed-arms",
      "suiteId": "productivity",
      "performance": {
        "prefillTokensPerSec": 501,
        "decodeTokensPerSec": 36.7,
        "samples": 2
      },
      "runtime": {
        "contextTokens": 262144,
        "peakMemoryMb": 44772,
        "kvCacheType": "f16"
      },
      "cells": [
        {
          "scenarioId": "constrained-comms",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 62495
        },
        {
          "scenarioId": "records-intake",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 625868
        },
        {
          "scenarioId": "plan-and-estimate",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 147854
        },
        {
          "scenarioId": "meeting-followup",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 158326
        },
        {
          "scenarioId": "craftbook-spreadsheet-model",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 944247
        },
        {
          "scenarioId": "conflict-synthesis",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 389269
        },
        {
          "scenarioId": "docblocks-theme-roundtrip",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 102956
        },
        {
          "scenarioId": "craftbook-powerpoint-deck",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 705815
        },
        {
          "scenarioId": "wikipedia-research-brief",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 675472
        },
        {
          "scenarioId": "craftbook-week-plan",
          "trials": 3,
          "successes": 1,
          "nonModelFailures": 0,
          "medianDurationMs": 5106543
        },
        {
          "scenarioId": "craftbook-ab-test-readout",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 1860343
        },
        {
          "scenarioId": "craftbook-annotated-bibliography",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 506547
        },
        {
          "scenarioId": "craftbook-research-to-document",
          "trials": 3,
          "successes": 3,
          "nonModelFailures": 0,
          "medianDurationMs": 497444
        }
      ]
    }
  ]
}
