[
 {
  "pass": 44.0,
  "score": 75.9,
  "micro": 71.5,
  "macro": 71.2,
  "clean": 9.1,
  "link": 22.7,
  "abs": 34.9,
  "layered": 73.5,
  "in_k": 569.4,
  "out_k": 9.5,
  "total_k": 578.8,
  "cost": 0.436,
  "model": "All models",
  "harness": "Claude Code"
 },
 {
  "pass": 35.8,
  "score": 70.7,
  "micro": 60.2,
  "macro": 59.8,
  "clean": 2.8,
  "link": 28.1,
  "abs": 33.0,
  "layered": 65.2,
  "in_k": 606.9,
  "out_k": 10.2,
  "total_k": 617.2,
  "cost": 0.595,
  "model": "All models",
  "harness": "Codex"
 },
 {
  "pass": 40.0,
  "score": 73.3,
  "micro": 69.5,
  "macro": 68.8,
  "clean": 9.8,
  "link": 21.8,
  "abs": 30.2,
  "layered": 71.0,
  "in_k": 361.5,
  "out_k": 11.2,
  "total_k": 372.7,
  "cost": 0.446,
  "model": "All models",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 38.2,
  "score": 71.7,
  "micro": 71.9,
  "macro": 72.0,
  "clean": 11.7,
  "link": 26.0,
  "abs": 26.5,
  "layered": 71.8,
  "in_k": 670.8,
  "out_k": 9.2,
  "total_k": 680.0,
  "cost": 0.633,
  "model": "All models",
  "harness": "OpenClaw"
 },
 {
  "pass": 39.5,
  "score": 72.9,
  "micro": 68.3,
  "macro": 67.9,
  "clean": 8.4,
  "link": 24.6,
  "abs": 31.1,
  "layered": 70.4,
  "in_k": 552.1,
  "out_k": 10.0,
  "total_k": 562.1,
  "cost": 0.528,
  "model": "All models",
  "harness": "All Harnesses"
 },
 {
  "pass": 43.2,
  "score": 72.5,
  "micro": 71.6,
  "macro": 71.2,
  "clean": 8.4,
  "link": 22.1,
  "abs": 34.7,
  "layered": 71.9,
  "in_k": 603.5,
  "out_k": 12.9,
  "total_k": 616.5,
  "cost": 0.115,
  "model": "DeepSeek-V4-Pro",
  "harness": "Claude Code"
 },
 {
  "pass": 30.5,
  "score": 68.4,
  "micro": 59.9,
  "macro": 59.3,
  "clean": 0.0,
  "link": 27.4,
  "abs": 30.5,
  "layered": 63.8,
  "in_k": 580.1,
  "out_k": 13.0,
  "total_k": 593.1,
  "cost": 0.046,
  "model": "DeepSeek-V4-Pro",
  "harness": "Codex"
 },
 {
  "pass": 37.9,
  "score": 69.9,
  "micro": 68.6,
  "macro": 67.9,
  "clean": 10.5,
  "link": 25.3,
  "abs": 27.4,
  "layered": 68.9,
  "in_k": 325.0,
  "out_k": 11.4,
  "total_k": 336.4,
  "cost": 0.043,
  "model": "DeepSeek-V4-Pro",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 38.9,
  "score": 72.0,
  "micro": 71.9,
  "macro": 71.9,
  "clean": 11.6,
  "link": 25.3,
  "abs": 27.4,
  "layered": 72.0,
  "in_k": 605.0,
  "out_k": 10.8,
  "total_k": 615.8,
  "cost": 0.094,
  "model": "DeepSeek-V4-Pro",
  "harness": "OpenClaw"
 },
 {
  "pass": 40.0,
  "score": 74.1,
  "micro": 70.4,
  "macro": 69.7,
  "clean": 8.4,
  "link": 21.1,
  "abs": 31.6,
  "layered": 71.9,
  "in_k": 590.7,
  "out_k": 10.3,
  "total_k": 601.0,
  "cost": 0.247,
  "model": "Gemini 3.7 Flash",
  "harness": "Claude Code"
 },
 {
  "pass": 38.9,
  "score": 73.8,
  "micro": 62.5,
  "macro": 62.0,
  "clean": 7.4,
  "link": 23.2,
  "abs": 31.6,
  "layered": 67.9,
  "in_k": 622.2,
  "out_k": 13.1,
  "total_k": 635.3,
  "cost": 0.226,
  "model": "Gemini 3.7 Flash",
  "harness": "Codex"
 },
 {
  "pass": 44.2,
  "score": 74.6,
  "micro": 70.8,
  "macro": 70.0,
  "clean": 8.4,
  "link": 25.3,
  "abs": 35.8,
  "layered": 72.3,
  "in_k": 496.0,
  "out_k": 11.6,
  "total_k": 507.6,
  "cost": 0.416,
  "model": "Gemini 3.7 Flash",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 38.9,
  "score": 74.2,
  "micro": 75.1,
  "macro": 74.8,
  "clean": 10.5,
  "link": 26.3,
  "abs": 28.4,
  "layered": 74.5,
  "in_k": 975.9,
  "out_k": 10.6,
  "total_k": 986.5,
  "cost": 0.772,
  "model": "Gemini 3.7 Flash",
  "harness": "OpenClaw"
 },
 {
  "pass": 43.2,
  "score": 76.0,
  "micro": 73.7,
  "macro": 73.1,
  "clean": 9.5,
  "link": 21.1,
  "abs": 33.7,
  "layered": 74.6,
  "in_k": 639.3,
  "out_k": 13.9,
  "total_k": 653.2,
  "cost": 0.408,
  "model": "GLM-5.3",
  "harness": "Claude Code"
 },
 {
  "pass": 28.4,
  "score": 68.6,
  "micro": 59.1,
  "macro": 59.3,
  "clean": 2.1,
  "link": 28.4,
  "abs": 26.3,
  "layered": 63.9,
  "in_k": 627.1,
  "out_k": 8.2,
  "total_k": 635.3,
  "cost": 0.189,
  "model": "GLM-5.3",
  "harness": "Codex"
 },
 {
  "pass": 44.2,
  "score": 75.5,
  "micro": 71.4,
  "macro": 70.5,
  "clean": 9.5,
  "link": 17.9,
  "abs": 34.7,
  "layered": 73.0,
  "in_k": 380.6,
  "out_k": 14.5,
  "total_k": 395.1,
  "cost": 0.226,
  "model": "GLM-5.3",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 41.1,
  "score": 67.5,
  "micro": 69.7,
  "macro": 70.1,
  "clean": 12.6,
  "link": 25.3,
  "abs": 28.4,
  "layered": 68.8,
  "in_k": 606.1,
  "out_k": 10.2,
  "total_k": 616.3,
  "cost": 0.18,
  "model": "GLM-5.3",
  "harness": "OpenClaw"
 },
 {
  "pass": 44.2,
  "score": 79.6,
  "micro": 72.8,
  "macro": 72.6,
  "clean": 11.6,
  "link": 21.1,
  "abs": 32.6,
  "layered": 76.1,
  "in_k": 559.5,
  "out_k": 4.7,
  "total_k": 564.2,
  "cost": 0.814,
  "model": "GPT-5.6-Sol",
  "harness": "Claude Code"
 },
 {
  "pass": 41.1,
  "score": 71.8,
  "micro": 57.6,
  "macro": 56.4,
  "clean": 3.2,
  "link": 25.3,
  "abs": 37.9,
  "layered": 64.1,
  "in_k": 1427.8,
  "out_k": 10.9,
  "total_k": 1438.7,
  "cost": 1.649,
  "model": "GPT-5.6-Sol",
  "harness": "Codex"
 },
 {
  "pass": 36.8,
  "score": 73.2,
  "micro": 71.3,
  "macro": 70.4,
  "clean": 11.6,
  "link": 22.1,
  "abs": 25.3,
  "layered": 71.8,
  "in_k": 298.3,
  "out_k": 2.6,
  "total_k": 300.9,
  "cost": 0.361,
  "model": "GPT-5.6-Sol",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 42.1,
  "score": 71.5,
  "micro": 71.1,
  "macro": 71.2,
  "clean": 14.7,
  "link": 29.5,
  "abs": 27.4,
  "layered": 71.4,
  "in_k": 606.7,
  "out_k": 5.1,
  "total_k": 611.8,
  "cost": 0.974,
  "model": "GPT-5.6-Sol",
  "harness": "OpenClaw"
 },
 {
  "pass": 55.8,
  "score": 81.6,
  "micro": 71.9,
  "macro": 71.9,
  "clean": 10.5,
  "link": 18.9,
  "abs": 45.3,
  "layered": 76.7,
  "in_k": 531.1,
  "out_k": 7.2,
  "total_k": 538.3,
  "cost": 0.509,
  "model": "Grok-4.6",
  "harness": "Claude Code"
 },
 {
  "pass": 44.2,
  "score": 72.9,
  "micro": 59.6,
  "macro": 59.5,
  "clean": 5.3,
  "link": 21.1,
  "abs": 38.9,
  "layered": 66.2,
  "in_k": 292.6,
  "out_k": 4.2,
  "total_k": 296.9,
  "cost": 0.462,
  "model": "Grok-4.6",
  "harness": "Codex"
 },
 {
  "pass": 43.2,
  "score": 75.6,
  "micro": 71.4,
  "macro": 70.6,
  "clean": 10.5,
  "link": 18.9,
  "abs": 32.6,
  "layered": 73.1,
  "in_k": 315.7,
  "out_k": 11.9,
  "total_k": 327.6,
  "cost": 0.378,
  "model": "Grok-4.6",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 33.7,
  "score": 70.8,
  "micro": 68.5,
  "macro": 69.0,
  "clean": 8.4,
  "link": 24.2,
  "abs": 25.3,
  "layered": 69.9,
  "in_k": 541.6,
  "out_k": 4.4,
  "total_k": 546.0,
  "cost": 0.822,
  "model": "Grok-4.6",
  "harness": "OpenClaw"
 },
 {
  "pass": 32.6,
  "score": 69.5,
  "micro": 69.7,
  "macro": 69.2,
  "clean": 6.3,
  "link": 30.5,
  "abs": 26.3,
  "layered": 69.3,
  "in_k": 798.9,
  "out_k": 15.0,
  "total_k": 813.9,
  "cost": 0.044,
  "model": "HY3",
  "harness": "Claude Code"
 },
 {
  "pass": 30.5,
  "score": 64.2,
  "micro": 59.3,
  "macro": 59.0,
  "clean": 1.1,
  "link": 35.8,
  "abs": 29.5,
  "layered": 61.6,
  "in_k": 415.9,
  "out_k": 13.7,
  "total_k": 429.6,
  "cost": 0.026,
  "model": "HY3",
  "harness": "Codex"
 },
 {
  "pass": 28.4,
  "score": 64.8,
  "micro": 65.8,
  "macro": 65.7,
  "clean": 6.3,
  "link": 30.5,
  "abs": 22.1,
  "layered": 65.2,
  "in_k": 247.0,
  "out_k": 11.2,
  "total_k": 258.2,
  "cost": 0.018,
  "model": "HY3",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 29.5,
  "score": 70.0,
  "micro": 72.7,
  "macro": 72.7,
  "clean": 11.6,
  "link": 29.5,
  "abs": 17.9,
  "layered": 71.4,
  "in_k": 659.5,
  "out_k": 18.5,
  "total_k": 678.0,
  "cost": 0.041,
  "model": "HY3",
  "harness": "OpenClaw"
 },
 {
  "pass": 49.5,
  "score": 78.2,
  "micro": 71.1,
  "macro": 70.6,
  "clean": 11.6,
  "link": 21.1,
  "abs": 37.9,
  "layered": 74.4,
  "in_k": 358.1,
  "out_k": 4.8,
  "total_k": 362.9,
  "cost": 0.613,
  "model": "Kimi K3",
  "harness": "Claude Code"
 },
 {
  "pass": 42.1,
  "score": 74.0,
  "micro": 61.3,
  "macro": 60.9,
  "clean": 2.1,
  "link": 28.4,
  "abs": 40.0,
  "layered": 67.4,
  "in_k": 573.9,
  "out_k": 7.7,
  "total_k": 581.6,
  "cost": 0.379,
  "model": "Kimi K3",
  "harness": "Codex"
 },
 {
  "pass": 43.2,
  "score": 74.4,
  "micro": 71.5,
  "macro": 70.5,
  "clean": 10.5,
  "link": 16.8,
  "abs": 32.6,
  "layered": 72.4,
  "in_k": 281.7,
  "out_k": 6.1,
  "total_k": 287.8,
  "cost": 0.225,
  "model": "Kimi K3",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 40.0,
  "score": 73.8,
  "micro": 73.6,
  "macro": 73.5,
  "clean": 10.5,
  "link": 22.1,
  "abs": 29.5,
  "layered": 73.7,
  "in_k": 580.8,
  "out_k": 5.7,
  "total_k": 586.6,
  "cost": 0.351,
  "model": "Kimi K3",
  "harness": "OpenClaw"
 },
 {
  "pass": 53.7,
  "score": 83.4,
  "micro": 71.8,
  "macro": 71.9,
  "clean": 7.4,
  "link": 16.8,
  "abs": 46.3,
  "layered": 77.6,
  "in_k": 379.4,
  "out_k": 3.7,
  "total_k": 383.1,
  "cost": 0.66,
  "model": "Opus 5",
  "harness": "Claude Code"
 },
 {
  "pass": 30.5,
  "score": 70.3,
  "micro": 60.8,
  "macro": 60.5,
  "clean": 2.1,
  "link": 34.7,
  "abs": 28.4,
  "layered": 65.4,
  "in_k": 482.8,
  "out_k": 7.6,
  "total_k": 490.4,
  "cost": 2.121,
  "model": "Opus 5",
  "harness": "Codex"
 },
 {
  "pass": 49.5,
  "score": 79.3,
  "micro": 72.3,
  "macro": 71.4,
  "clean": 13.7,
  "link": 17.9,
  "abs": 35.8,
  "layered": 75.4,
  "in_k": 551.5,
  "out_k": 16.9,
  "total_k": 568.5,
  "cost": 2.071,
  "model": "Opus 5",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 41.1,
  "score": 76.1,
  "micro": 73.8,
  "macro": 73.9,
  "clean": 11.6,
  "link": 23.2,
  "abs": 29.5,
  "layered": 75.0,
  "in_k": 822.3,
  "out_k": 7.7,
  "total_k": 830.0,
  "cost": 2.168,
  "model": "Opus 5",
  "harness": "OpenClaw"
 },
 {
  "pass": 33.7,
  "score": 68.3,
  "micro": 70.5,
  "macro": 70.2,
  "clean": 8.4,
  "link": 31.6,
  "abs": 25.3,
  "layered": 69.3,
  "in_k": 663.8,
  "out_k": 12.6,
  "total_k": 676.4,
  "cost": 0.517,
  "model": "Qwen3.8-Max",
  "harness": "Claude Code"
 },
 {
  "pass": 35.8,
  "score": 72.1,
  "micro": 61.4,
  "macro": 61.0,
  "clean": 2.1,
  "link": 28.4,
  "abs": 33.7,
  "layered": 66.5,
  "in_k": 436.7,
  "out_k": 13.4,
  "total_k": 450.2,
  "cost": 0.258,
  "model": "Qwen3.8-Max",
  "harness": "Codex"
 },
 {
  "pass": 32.6,
  "score": 71.8,
  "micro": 62.5,
  "macro": 62.6,
  "clean": 7.4,
  "link": 21.1,
  "abs": 25.3,
  "layered": 67.2,
  "in_k": 357.9,
  "out_k": 14.2,
  "total_k": 372.1,
  "cost": 0.279,
  "model": "Qwen3.8-Max",
  "harness": "DeepSeek Harness"
 },
 {
  "pass": 38.9,
  "score": 69.0,
  "micro": 70.6,
  "macro": 70.9,
  "clean": 13.7,
  "link": 28.4,
  "abs": 25.3,
  "layered": 70.0,
  "in_k": 638.8,
  "out_k": 9.7,
  "total_k": 648.5,
  "cost": 0.29,
  "model": "Qwen3.8-Max",
  "harness": "OpenClaw"
 }
]