{
 "schema_version": 4,
 "generated_at": "2026-10-13T21:04:32+00:00",
 "title": "Multi-agent tasks: which coding agent",
 "how_to_use": "For task class X use classes.X.first_choice; if it cannot run right now (plan limit reached, engine unavailable), take the next entry in fallbacks. For coding and agentic work, use classes.X.subclasses.Y instead when the task is clearly kind Y.",
 "weights": {
  "public": 0.5,
  "own_success": 0.3,
  "speed": 0.2
 },
 "method": {
  "formula": "priority = 0.5*public_norm + 0.3*own_success + 0.2*speed_rank",
  "tie_band": 0.03,
  "tie_order": [
   "own_jobs desc",
   "speed_rank desc"
  ],
  "evidence_rule": "An unproven entry (no own jobs) ranks above a proven one only if its priority is more than 0.03 higher and its public_norm is higher.",
  "harness_discount": 0.8,
  "harness_rule": "A public metric measured in another agent's harness counts at 0.8x (norm multiplied). The engine's own harness, or a model-only benchmark (no harness or a neutral one), counts in full.",
  "neutral_harnesses": [
   "mini-swe-agent",
   "swe-agent",
   "terminus",
   "terminus 2",
   "terminus-2"
  ],
  "min_jobs_for_own_success": 10,
  "own_outcomes_window_days": 28,
  "subclasses": "Coding and agentic subclasses are ranked the same way, on the public metrics that measure that kind of work and own jobs of that kind; a subclass with neither is not listed.",
  "limits": "Plan limits are checked live when a job is dispatched; this table does not track them.",
  "status_meanings": {
   "proven": "ranked with own jobs from the last 28 days",
   "unproven": "public scores only, no own jobs yet"
  }
 },
 "engines": {
  "claude": {
   "display": "Claude Code",
   "cli": "claude",
   "models": [
    "claude-fable-5.1",
    "claude-opus-5.5",
    "claude-sonnet-5.5",
    "claude-haiku-5.5"
   ],
   "models_verified": true,
   "not_listed": []
  },
  "codex": {
   "display": "Codex CLI",
   "cli": "codex",
   "models": [
    "gpt-6.1-sol",
    "gpt-6-astra",
    "gpt-6-sol",
    "gpt-6-luna",
    "gpt-5.6-sol",
    "gpt-5.6-terra",
    "gpt-5.6-luna"
   ],
   "models_verified": true,
   "not_listed": []
  },
  "agy": {
   "display": "Antigravity CLI",
   "cli": "agy",
   "models": [
    "gemini-3.8-flash",
    "claude-opus-5.5",
    "claude-sonnet-5.5",
    "gpt-oss-120b"
   ],
   "models_verified": true,
   "not_listed": [
    "gemini-4-argon"
   ]
  },
  "grok": {
   "display": "Grok CLI",
   "cli": "grok",
   "models": [
    "grok-4.7",
    "grok-4.7-build-fast",
    "grok-4.6",
    "grok-4.5"
   ],
   "models_verified": false,
   "not_listed": []
  }
 },
 "classes": {
  "coding": {
   "description": "repo-level code changes",
   "metrics_used": [
    "swe_bench_verified_resolved",
    "swe_bench_multilingual_resolved",
    "terminal_bench_4_accuracy",
    "aa_coding"
   ],
   "first_choice": {
    "engine": "claude",
    "engine_name": "Claude Code",
    "model": "claude-opus-5.5",
    "model_display": "Claude Opus 5.5",
    "priority": 0.96,
    "status": [
     "proven"
    ],
    "rank_rule": null,
    "tie_broken": false,
    "inputs": {
     "public_norm": 1.0,
     "public_default": false,
     "public_metrics": {
      "terminal_bench_4_accuracy": {
       "score": 64.85,
       "norm": 1.0,
       "norm_before_discount": 1.0,
       "harness_relation": "own",
       "harness_factor": 1.0,
       "leaderboard_name": "Opus 5.5 (max)",
       "harness": "Claude Code",
       "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-opus-5-5-max-claude-code.json"
      }
     },
     "own_success": 1.0,
     "own_jobs": 15,
     "own_success_rate": 1.0,
     "own_success_default": false,
     "own_median_minutes": 1.9,
     "speed_rank": 0.8,
     "speed_source": "own median minutes"
    }
   },
   "fallbacks": [
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-fable-5.1",
     "model_display": "Claude Fable 5.1",
     "priority": 0.8012,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.867,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 57.88,
        "norm": 0.867,
        "norm_before_discount": 0.867,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "Fable 5.1 (max)",
        "harness": "Claude Code",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-02-anthropic-claude-fable-5-1-max-claude-code.json"
       }
      },
      "own_success": 0.959,
      "own_jobs": 73,
      "own_success_rate": 0.959,
      "own_success_default": false,
      "own_median_minutes": 2.9,
      "speed_rank": 0.4,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-sonnet-5.5",
     "model_display": "Claude Sonnet 5.5",
     "priority": 0.7861,
     "status": [
      "proven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.942,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 61.82,
        "norm": 0.942,
        "norm_before_discount": 0.942,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "Sonnet 5.5 (max)",
        "harness": "Claude Code",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-sonnet-5-5-max-claude-code.json"
       }
      },
      "own_success": 0.917,
      "own_jobs": 48,
      "own_success_rate": 0.917,
      "own_success_default": false,
      "own_median_minutes": 4.0,
      "speed_rank": 0.2,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-sol",
     "model_display": "GPT-6 Sol",
     "priority": 0.7025,
     "status": [
      "proven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.705,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 49.39,
        "norm": 0.705,
        "norm_before_discount": 0.705,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-6 Sol (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-sol-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 1,
      "own_success_rate": 1.0,
      "own_success_default": true,
      "own_median_minutes": 1.8,
      "speed_rank": 1.0,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6.1-sol",
     "model_display": "GPT-6.1 Sol",
     "priority": 0.6576,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.873,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 58.18,
        "norm": 0.873,
        "norm_before_discount": 0.873,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-6.1 Sol (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-1-sol-max-codex.json"
       }
      },
      "own_success": 0.737,
      "own_jobs": 19,
      "own_success_rate": 0.737,
      "own_success_default": false,
      "own_median_minutes": 10.2,
      "speed_rank": 0.0,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "claude-opus-5.5",
     "model_display": "Claude Opus 5.5",
     "priority": 0.65,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.8,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 64.85,
        "norm": 0.8,
        "norm_before_discount": 1.0,
        "harness_relation": "other",
        "harness_factor": 0.8,
        "leaderboard_name": "Opus 5.5 (max)",
        "harness": "Claude Code",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-opus-5-5-max-claude-code.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "claude-sonnet-5.5",
     "model_display": "Claude Sonnet 5.5",
     "priority": 0.627,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.754,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 61.82,
        "norm": 0.754,
        "norm_before_discount": 0.942,
        "harness_relation": "other",
        "harness_factor": 0.8,
        "leaderboard_name": "Sonnet 5.5 (max)",
        "harness": "Claude Code",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-sonnet-5-5-max-claude-code.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-astra",
     "model_display": "GPT-6 Astra",
     "priority": 0.52,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.5,
      "own_jobs": 4,
      "own_success_rate": 1.0,
      "own_success_default": true,
      "own_median_minutes": 2.1,
      "speed_rank": 0.6,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "gpt-oss-120b",
     "model_display": "GPT-OSS 120B",
     "priority": 0.5,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.5,
      "public_default": false,
      "public_metrics": {
       "swe_bench_verified_resolved": {
        "score": 26.0,
        "norm": 0.5,
        "norm_before_discount": 0.5,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-oss-120b",
        "harness": "mini-SWE-agent",
        "source_url": "https://www.swebench.com/"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-sol",
     "model_display": "GPT-5.6 Sol",
     "priority": 0.487,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.474,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 37.27,
        "norm": 0.474,
        "norm_before_discount": 0.474,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-5.6 Sol (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-sol-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-terra",
     "model_display": "GPT-5.6 Terra",
     "priority": 0.337,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.174,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 21.52,
        "norm": 0.174,
        "norm_before_discount": 0.174,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-5.6 Terra (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-terra-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.6",
     "model_display": "Grok 4.6",
     "priority": 0.325,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.15,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 20.3,
        "norm": 0.15,
        "norm_before_discount": 0.15,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "Grok 4.6 (none)",
        "harness": "Grok Build",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-xai-grok-4-6-none-grok-build.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "gemini-3.8-flash",
     "model_display": "Gemini 3.8 Flash",
     "priority": 0.3135,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.127,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 19.09,
        "norm": 0.127,
        "norm_before_discount": 0.127,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "Gemini 3.8 Flash (high)",
        "harness": "mini-SWE-agent",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-01-gemini-gemini-3-8-flash-high-mini-swe-agent.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-luna",
     "model_display": "GPT-5.6 Luna",
     "priority": 0.2965,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.093,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 17.27,
        "norm": 0.093,
        "norm_before_discount": 0.093,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-5.6 Luna (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-luna-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-luna",
     "model_display": "GPT-6 Luna",
     "priority": 0.2875,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.075,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 16.36,
        "norm": 0.075,
        "norm_before_discount": 0.075,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-6 Luna (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-luna-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.5",
     "model_display": "Grok 4.5",
     "priority": 0.25,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.0,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 12.42,
        "norm": 0.0,
        "norm_before_discount": 0.0,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "Grok 4.5 (none)",
        "harness": "Grok Build",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-xai-grok-4-5-none-grok-build.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    }
   ],
   "subclasses": {
    "bugfix": {
     "description": "finding and fixing defects in existing code",
     "metrics_used": [
      "swe_bench_verified_resolved"
     ],
     "first_choice": {
      "engine": "claude",
      "engine_name": "Claude Code",
      "model": "claude-sonnet-5.5",
      "model_display": "Claude Sonnet 5.5",
      "priority": 0.65,
      "status": [
       "proven"
      ],
      "rank_rule": null,
      "tie_broken": false,
      "inputs": {
       "public_norm": 0.5,
       "public_default": true,
       "public_metrics": {},
       "own_success": 1.0,
       "own_jobs": 13,
       "own_success_rate": 1.0,
       "own_success_default": false,
       "own_median_minutes": 3.3,
       "speed_rank": 0.5,
       "speed_source": "own median minutes"
      }
     },
     "fallbacks": [
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "gpt-oss-120b",
       "model_display": "GPT-OSS 120B",
       "priority": 0.5,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.5,
        "public_default": false,
        "public_metrics": {
         "swe_bench_verified_resolved": {
          "score": 26.0,
          "norm": 0.5,
          "norm_before_discount": 0.5,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-oss-120b",
          "harness": "mini-SWE-agent",
          "source_url": "https://www.swebench.com/"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      }
     ]
    },
    "features": {
     "description": "new features, tools and restructuring",
     "metrics_used": [
      "arena_text_coding"
     ],
     "first_choice": {
      "engine": "claude",
      "engine_name": "Claude Code",
      "model": "claude-sonnet-5.5",
      "model_display": "Claude Sonnet 5.5",
      "priority": 0.899,
      "status": [
       "proven"
      ],
      "rank_rule": null,
      "tie_broken": false,
      "inputs": {
       "public_norm": 0.798,
       "public_default": false,
       "public_metrics": {
        "arena_text_coding": {
         "score": 1513.3108959838664,
         "norm": 0.798,
         "norm_before_discount": 0.798,
         "harness_relation": "model-only",
         "harness_factor": 1.0,
         "leaderboard_name": "claude-sonnet-5.5-xhigh",
         "harness": "",
         "source_url": "https://arena.ai/leaderboard/text"
        }
       },
       "own_success": 1.0,
       "own_jobs": 17,
       "own_success_rate": 1.0,
       "own_success_default": false,
       "own_median_minutes": 5.5,
       "speed_rank": 1.0,
       "speed_source": "own median minutes"
      }
     },
     "fallbacks": [
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-opus-5.5",
       "model_display": "Claude Opus 5.5",
       "priority": 0.65,
       "status": [
        "proven"
       ],
       "rank_rule": "evidence",
       "tie_broken": false,
       "inputs": {
        "public_norm": 1.0,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1547.239145968882,
          "norm": 1.0,
          "norm_before_discount": 1.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-opus-5.5-high",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 2,
        "own_success_rate": 1.0,
        "own_success_default": true,
        "own_median_minutes": 9.0,
        "speed_rank": 0.0,
        "speed_source": "own median minutes"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "claude-opus-5.5",
       "model_display": "Claude Opus 5.5",
       "priority": 0.75,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 1.0,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1547.239145968882,
          "norm": 1.0,
          "norm_before_discount": 1.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-opus-5.5-high",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-fable-5.1",
       "model_display": "Claude Fable 5.1",
       "priority": 0.6915,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.883,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1527.5846906589086,
          "norm": 0.883,
          "norm_before_discount": 0.883,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-fable-5.1-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "claude-sonnet-5.5",
       "model_display": "Claude Sonnet 5.5",
       "priority": 0.649,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.798,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1513.3108959838664,
          "norm": 0.798,
          "norm_before_discount": 0.798,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-sonnet-5.5-xhigh",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "gemini-3.8-flash",
       "model_display": "Gemini 3.8 Flash",
       "priority": 0.6385,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.777,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1509.8711462941833,
          "norm": 0.777,
          "norm_before_discount": 0.777,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gemini-3.8-flash-high",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-sol",
       "model_display": "GPT-5.6 Sol",
       "priority": 0.602,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.704,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1497.5233773583282,
          "norm": 0.704,
          "norm_before_discount": 0.704,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-5.6-sol-xhigh",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-astra",
       "model_display": "GPT-6 Astra",
       "priority": 0.571,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.642,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1487.1260770764898,
          "norm": 0.642,
          "norm_before_discount": 0.642,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-6-astra-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6.1-sol",
       "model_display": "GPT-6.1 Sol",
       "priority": 0.5705,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.641,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1486.975300059475,
          "norm": 0.641,
          "norm_before_discount": 0.641,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-6.1-sol-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-terra",
       "model_display": "GPT-5.6 Terra",
       "priority": 0.56,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.62,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1483.5614136879308,
          "norm": 0.62,
          "norm_before_discount": 0.62,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-5.6-terra-xhigh",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.5",
       "model_display": "Grok 4.5",
       "priority": 0.531,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.562,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1473.7903598409762,
          "norm": 0.562,
          "norm_before_discount": 0.562,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "grok-4.5",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-luna",
       "model_display": "GPT-5.6 Luna",
       "priority": 0.5075,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.515,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1465.9273539810993,
          "norm": 0.515,
          "norm_before_discount": 0.515,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-5.6-luna-xhigh",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.6",
       "model_display": "Grok 4.6",
       "priority": 0.5035,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.507,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1464.5162259520284,
          "norm": 0.507,
          "norm_before_discount": 0.507,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "grok-4.6-high",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-sol",
       "model_display": "GPT-6 Sol",
       "priority": 0.4525,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.405,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1447.4141051086413,
          "norm": 0.405,
          "norm_before_discount": 0.405,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-6-sol-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-luna",
       "model_display": "GPT-6 Luna",
       "priority": 0.4285,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.357,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1439.3970542539028,
          "norm": 0.357,
          "norm_before_discount": 0.357,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-6-luna-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.7",
       "model_display": "Grok 4.7",
       "priority": 0.3905,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.281,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1426.6393373796104,
          "norm": 0.281,
          "norm_before_discount": 0.281,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "grok-4.7-xhigh",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "gpt-oss-120b",
       "model_display": "GPT-OSS 120B",
       "priority": 0.25,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.0,
        "public_default": false,
        "public_metrics": {
         "arena_text_coding": {
          "score": 1379.4836027510644,
          "norm": 0.0,
          "norm_before_discount": 0.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-oss-120b",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/text"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      }
     ]
    },
    "frontend": {
     "description": "web pages, UI components and styling",
     "metrics_used": [
      "arena_webdev_overall"
     ],
     "first_choice": {
      "engine": "agy",
      "engine_name": "Antigravity CLI",
      "model": "claude-opus-5.5",
      "model_display": "Claude Opus 5.5",
      "priority": 0.75,
      "status": [
       "unproven"
      ],
      "rank_rule": "tie",
      "tie_broken": true,
      "inputs": {
       "public_norm": 1.0,
       "public_default": false,
       "public_metrics": {
        "arena_webdev_overall": {
         "score": 1813.028149728328,
         "norm": 1.0,
         "norm_before_discount": 1.0,
         "harness_relation": "model-only",
         "harness_factor": 1.0,
         "leaderboard_name": "claude-opus-5.5-max",
         "harness": "",
         "source_url": "https://arena.ai/leaderboard/webdev"
        }
       },
       "own_success": 0.5,
       "own_jobs": 0,
       "own_success_rate": null,
       "own_success_default": true,
       "own_median_minutes": null,
       "speed_rank": 0.5,
       "speed_source": "default"
      }
     },
     "fallbacks": [
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-opus-5.5",
       "model_display": "Claude Opus 5.5",
       "priority": 0.75,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 1.0,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1813.028149728328,
          "norm": 1.0,
          "norm_before_discount": 1.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-opus-5.5-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-sonnet-5.5",
       "model_display": "Claude Sonnet 5.5",
       "priority": 0.6755,
       "status": [
        "proven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.851,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1774.2855361868098,
          "norm": 0.851,
          "norm_before_discount": 0.851,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-sonnet-5.5-xhigh",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 1,
        "own_success_rate": 1.0,
        "own_success_default": true,
        "own_median_minutes": 2.8,
        "speed_rank": 0.5,
        "speed_source": "own median minutes"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-astra",
       "model_display": "GPT-6 Astra",
       "priority": 0.6985,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.897,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1786.1141116447413,
          "norm": 0.897,
          "norm_before_discount": 0.897,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-6-astra-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "claude-sonnet-5.5",
       "model_display": "Claude Sonnet 5.5",
       "priority": 0.6755,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.851,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1774.2855361868098,
          "norm": 0.851,
          "norm_before_discount": 0.851,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-sonnet-5.5-xhigh",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6.1-sol",
       "model_display": "GPT-6.1 Sol",
       "priority": 0.6395,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.779,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1755.3982673326393,
          "norm": 0.779,
          "norm_before_discount": 0.779,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-6.1-sol-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-fable-5.1",
       "model_display": "Claude Fable 5.1",
       "priority": 0.617,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.734,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1743.7093865265408,
          "norm": 0.734,
          "norm_before_discount": 0.734,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-fable-5.1-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-sol",
       "model_display": "GPT-6 Sol",
       "priority": 0.51,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.52,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1688.1815937715994,
          "norm": 0.52,
          "norm_before_discount": 0.52,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-6-sol-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.7",
       "model_display": "Grok 4.7",
       "priority": 0.416,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.332,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1639.090984165512,
          "norm": 0.332,
          "norm_before_discount": 0.332,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "grok-4.7-xhigh",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.6",
       "model_display": "Grok 4.6",
       "priority": 0.3735,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.247,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1617.1601717927338,
          "norm": 0.247,
          "norm_before_discount": 0.247,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "grok-4.6-high",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-haiku-5.5",
       "model_display": "Claude Haiku 5.5",
       "priority": 0.3155,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.131,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1586.9446989146388,
          "norm": 0.131,
          "norm_before_discount": 0.131,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "claude-haiku-5.5-high",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "gemini-3.8-flash",
       "model_display": "Gemini 3.8 Flash",
       "priority": 0.309,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.118,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1583.462059363355,
          "norm": 0.118,
          "norm_before_discount": 0.118,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gemini-3.8-flash-high",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-luna",
       "model_display": "GPT-6 Luna",
       "priority": 0.304,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.108,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1580.896017410467,
          "norm": 0.108,
          "norm_before_discount": 0.108,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "gpt-6-luna-max",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.5",
       "model_display": "Grok 4.5",
       "priority": 0.25,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.0,
        "public_default": false,
        "public_metrics": {
         "arena_webdev_overall": {
          "score": 1552.823224297653,
          "norm": 0.0,
          "norm_before_discount": 0.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "grok-4.5",
          "harness": "",
          "source_url": "https://arena.ai/leaderboard/webdev"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      }
     ]
    },
    "review": {
     "description": "code review, audits and tests",
     "metrics_used": [],
     "first_choice": {
      "engine": "claude",
      "engine_name": "Claude Code",
      "model": "claude-opus-5.5",
      "model_display": "Claude Opus 5.5",
      "priority": 0.75,
      "status": [
       "proven"
      ],
      "rank_rule": null,
      "tie_broken": false,
      "inputs": {
       "public_norm": 0.5,
       "public_default": true,
       "public_metrics": {},
       "own_success": 1.0,
       "own_jobs": 13,
       "own_success_rate": 1.0,
       "own_success_default": false,
       "own_median_minutes": 1.8,
       "speed_rank": 1.0,
       "speed_source": "own median minutes"
      }
     },
     "fallbacks": [
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-fable-5.1",
       "model_display": "Claude Fable 5.1",
       "priority": 0.5777,
       "status": [
        "proven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.5,
        "public_default": true,
        "public_metrics": {},
        "own_success": 0.959,
        "own_jobs": 73,
        "own_success_rate": 0.959,
        "own_success_default": false,
        "own_median_minutes": 2.9,
        "speed_rank": 0.2,
        "speed_source": "own median minutes"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-sonnet-5.5",
       "model_display": "Claude Sonnet 5.5",
       "priority": 0.5736,
       "status": [
        "proven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.5,
        "public_default": true,
        "public_metrics": {},
        "own_success": 0.812,
        "own_jobs": 16,
        "own_success_rate": 0.812,
        "own_success_default": false,
        "own_median_minutes": 2.8,
        "speed_rank": 0.4,
        "speed_source": "own median minutes"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-sol",
       "model_display": "GPT-6 Sol",
       "priority": 0.56,
       "status": [
        "proven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.5,
        "public_default": true,
        "public_metrics": {},
        "own_success": 0.5,
        "own_jobs": 1,
        "own_success_rate": 1.0,
        "own_success_default": true,
        "own_median_minutes": 1.8,
        "speed_rank": 0.8,
        "speed_source": "own median minutes"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-astra",
       "model_display": "GPT-6 Astra",
       "priority": 0.52,
       "status": [
        "proven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.5,
        "public_default": true,
        "public_metrics": {},
        "own_success": 0.5,
        "own_jobs": 2,
        "own_success_rate": 1.0,
        "own_success_default": true,
        "own_median_minutes": 2.1,
        "speed_rank": 0.6,
        "speed_source": "own median minutes"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6.1-sol",
       "model_display": "GPT-6.1 Sol",
       "priority": 0.4564,
       "status": [
        "proven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.5,
        "public_default": true,
        "public_metrics": {},
        "own_success": 0.688,
        "own_jobs": 16,
        "own_success_rate": 0.688,
        "own_success_default": false,
        "own_median_minutes": 10.9,
        "speed_rank": 0.0,
        "speed_source": "own median minutes"
       }
      }
     ]
    }
   }
  },
  "agentic": {
   "description": "terminal and multi-step tool work",
   "metrics_used": [
    "terminal_bench_4_accuracy",
    "aa_agentic"
   ],
   "first_choice": {
    "engine": "claude",
    "engine_name": "Claude Code",
    "model": "claude-opus-5.5",
    "model_display": "Claude Opus 5.5",
    "priority": 0.75,
    "status": [
     "unproven"
    ],
    "rank_rule": "tie",
    "tie_broken": true,
    "inputs": {
     "public_norm": 1.0,
     "public_default": false,
     "public_metrics": {
      "terminal_bench_4_accuracy": {
       "score": 64.85,
       "norm": 1.0,
       "norm_before_discount": 1.0,
       "harness_relation": "own",
       "harness_factor": 1.0,
       "leaderboard_name": "Opus 5.5 (max)",
       "harness": "Claude Code",
       "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-opus-5-5-max-claude-code.json"
      }
     },
     "own_success": 0.5,
     "own_jobs": 0,
     "own_success_rate": null,
     "own_success_default": true,
     "own_median_minutes": null,
     "speed_rank": 0.5,
     "speed_source": "default"
    }
   },
   "fallbacks": [
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-sonnet-5.5",
     "model_display": "Claude Sonnet 5.5",
     "priority": 0.721,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.942,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 61.82,
        "norm": 0.942,
        "norm_before_discount": 0.942,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "Sonnet 5.5 (max)",
        "harness": "Claude Code",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-sonnet-5-5-max-claude-code.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6.1-sol",
     "model_display": "GPT-6.1 Sol",
     "priority": 0.6865,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.873,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 58.18,
        "norm": 0.873,
        "norm_before_discount": 0.873,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-6.1 Sol (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-1-sol-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-fable-5.1",
     "model_display": "Claude Fable 5.1",
     "priority": 0.6835,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.867,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 57.88,
        "norm": 0.867,
        "norm_before_discount": 0.867,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "Fable 5.1 (max)",
        "harness": "Claude Code",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-02-anthropic-claude-fable-5-1-max-claude-code.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "claude-opus-5.5",
     "model_display": "Claude Opus 5.5",
     "priority": 0.65,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.8,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 64.85,
        "norm": 0.8,
        "norm_before_discount": 1.0,
        "harness_relation": "other",
        "harness_factor": 0.8,
        "leaderboard_name": "Opus 5.5 (max)",
        "harness": "Claude Code",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-opus-5-5-max-claude-code.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "claude-sonnet-5.5",
     "model_display": "Claude Sonnet 5.5",
     "priority": 0.627,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.754,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 61.82,
        "norm": 0.754,
        "norm_before_discount": 0.942,
        "harness_relation": "other",
        "harness_factor": 0.8,
        "leaderboard_name": "Sonnet 5.5 (max)",
        "harness": "Claude Code",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-sonnet-5-5-max-claude-code.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-sol",
     "model_display": "GPT-6 Sol",
     "priority": 0.6025,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.705,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 49.39,
        "norm": 0.705,
        "norm_before_discount": 0.705,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-6 Sol (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-sol-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-sol",
     "model_display": "GPT-5.6 Sol",
     "priority": 0.487,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.474,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 37.27,
        "norm": 0.474,
        "norm_before_discount": 0.474,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-5.6 Sol (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-sol-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-terra",
     "model_display": "GPT-5.6 Terra",
     "priority": 0.337,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.174,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 21.52,
        "norm": 0.174,
        "norm_before_discount": 0.174,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-5.6 Terra (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-terra-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.6",
     "model_display": "Grok 4.6",
     "priority": 0.325,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.15,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 20.3,
        "norm": 0.15,
        "norm_before_discount": 0.15,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "Grok 4.6 (none)",
        "harness": "Grok Build",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-xai-grok-4-6-none-grok-build.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "gemini-3.8-flash",
     "model_display": "Gemini 3.8 Flash",
     "priority": 0.3135,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.127,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 19.09,
        "norm": 0.127,
        "norm_before_discount": 0.127,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "Gemini 3.8 Flash (high)",
        "harness": "mini-SWE-agent",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-01-gemini-gemini-3-8-flash-high-mini-swe-agent.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-luna",
     "model_display": "GPT-5.6 Luna",
     "priority": 0.2965,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.093,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 17.27,
        "norm": 0.093,
        "norm_before_discount": 0.093,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-5.6 Luna (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-luna-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-luna",
     "model_display": "GPT-6 Luna",
     "priority": 0.2875,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.075,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 16.36,
        "norm": 0.075,
        "norm_before_discount": 0.075,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "GPT-6 Luna (max)",
        "harness": "Codex",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-luna-max-codex.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.5",
     "model_display": "Grok 4.5",
     "priority": 0.25,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.0,
      "public_default": false,
      "public_metrics": {
       "terminal_bench_4_accuracy": {
        "score": 12.42,
        "norm": 0.0,
        "norm_before_discount": 0.0,
        "harness_relation": "own",
        "harness_factor": 1.0,
        "leaderboard_name": "Grok 4.5 (none)",
        "harness": "Grok Build",
        "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-xai-grok-4-5-none-grok-build.json"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    }
   ],
   "subclasses": {
    "terminal": {
     "description": "shell, installs, configuration, server upkeep",
     "metrics_used": [
      "terminal_bench_4_accuracy",
      "arena_agent_bash_reliability_no_prereq_overall",
      "arena_agent_bash_recovery_steps_overall"
     ],
     "first_choice": {
      "engine": "claude",
      "engine_name": "Claude Code",
      "model": "claude-opus-5.5",
      "model_display": "Claude Opus 5.5",
      "priority": 0.7345,
      "status": [
       "unproven"
      ],
      "rank_rule": "tie",
      "tie_broken": true,
      "inputs": {
       "public_norm": 0.969,
       "public_default": false,
       "public_metrics": {
        "terminal_bench_4_accuracy": {
         "score": 64.85,
         "norm": 1.0,
         "norm_before_discount": 1.0,
         "harness_relation": "own",
         "harness_factor": 1.0,
         "leaderboard_name": "Opus 5.5 (max)",
         "harness": "Claude Code",
         "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-opus-5-5-max-claude-code.json"
        },
        "arena_agent_bash_reliability_no_prereq_overall": {
         "score": 0.1507932887447973,
         "norm": 1.0,
         "norm_before_discount": 1.0,
         "harness_relation": "model-only",
         "harness_factor": 1.0,
         "leaderboard_name": "Claude Opus 5.5 (Max)",
         "harness": "",
         "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
        },
        "arena_agent_bash_recovery_steps_overall": {
         "score": 0.1355335853279116,
         "norm": 0.906,
         "norm_before_discount": 0.906,
         "harness_relation": "model-only",
         "harness_factor": 1.0,
         "leaderboard_name": "Claude Opus 5.5 (High)",
         "harness": "",
         "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
        }
       },
       "own_success": 0.5,
       "own_jobs": 0,
       "own_success_rate": null,
       "own_success_default": true,
       "own_median_minutes": null,
       "speed_rank": 0.5,
       "speed_source": "default"
      }
     },
     "fallbacks": [
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-sonnet-5.5",
       "model_display": "Claude Sonnet 5.5",
       "priority": 0.7195,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.939,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 61.82,
          "norm": 0.942,
          "norm_before_discount": 0.942,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "Sonnet 5.5 (max)",
          "harness": "Claude Code",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-sonnet-5-5-max-claude-code.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.13048756603329853,
          "norm": 0.875,
          "norm_before_discount": 0.875,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Sonnet 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.15035238150522567,
          "norm": 1.0,
          "norm_before_discount": 1.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Sonnet 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "claude-opus-5.5",
       "model_display": "Claude Opus 5.5",
       "priority": 0.701,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.902,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 64.85,
          "norm": 0.8,
          "norm_before_discount": 1.0,
          "harness_relation": "other",
          "harness_factor": 0.8,
          "leaderboard_name": "Opus 5.5 (max)",
          "harness": "Claude Code",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-opus-5-5-max-claude-code.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.1507932887447973,
          "norm": 1.0,
          "norm_before_discount": 1.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Opus 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.1355335853279116,
          "norm": 0.906,
          "norm_before_discount": 0.906,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Opus 5.5 (High)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "claude-sonnet-5.5",
       "model_display": "Claude Sonnet 5.5",
       "priority": 0.688,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.876,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 61.82,
          "norm": 0.754,
          "norm_before_discount": 0.942,
          "harness_relation": "other",
          "harness_factor": 0.8,
          "leaderboard_name": "Sonnet 5.5 (max)",
          "harness": "Claude Code",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-anthropic-claude-sonnet-5-5-max-claude-code.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.13048756603329853,
          "norm": 0.875,
          "norm_before_discount": 0.875,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Sonnet 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.15035238150522567,
          "norm": 1.0,
          "norm_before_discount": 1.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Sonnet 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-fable-5.1",
       "model_display": "Claude Fable 5.1",
       "priority": 0.6465,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.793,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 57.88,
          "norm": 0.867,
          "norm_before_discount": 0.867,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "Fable 5.1 (max)",
          "harness": "Claude Code",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-02-anthropic-claude-fable-5-1-max-claude-code.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.10677538953017898,
          "norm": 0.73,
          "norm_before_discount": 0.73,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Fable 5.1 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.11603170489005432,
          "norm": 0.782,
          "norm_before_discount": 0.782,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Fable 5.1 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-sol",
       "model_display": "GPT-6 Sol",
       "priority": 0.5745,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.649,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 49.39,
          "norm": 0.705,
          "norm_before_discount": 0.705,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT-6 Sol (max)",
          "harness": "Codex",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-sol-max-codex.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.11033685871110352,
          "norm": 0.751,
          "norm_before_discount": 0.751,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Sol (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.07030190807524606,
          "norm": 0.491,
          "norm_before_discount": 0.491,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Sol (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-astra",
       "model_display": "GPT-6 Astra",
       "priority": 0.5355,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.571,
        "public_default": false,
        "public_metrics": {
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.10951229500284923,
          "norm": 0.746,
          "norm_before_discount": 0.746,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Astra (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.055274748105542004,
          "norm": 0.395,
          "norm_before_discount": 0.395,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Astra (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.7",
       "model_display": "Grok 4.7",
       "priority": 0.53,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.56,
        "public_default": false,
        "public_metrics": {
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.08876340407593719,
          "norm": 0.619,
          "norm_before_discount": 0.619,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.7 (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.07203833245634916,
          "norm": 0.502,
          "norm_before_discount": 0.502,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.7 (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6.1-sol",
       "model_display": "GPT-6.1 Sol",
       "priority": 0.49,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.48,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 58.18,
          "norm": 0.873,
          "norm_before_discount": 0.873,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT-6.1 Sol (max)",
          "harness": "Codex",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-1-sol-max-codex.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.026347442332075144,
          "norm": 0.236,
          "norm_before_discount": 0.236,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6.1 Sol (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.04528409239854958,
          "norm": 0.332,
          "norm_before_discount": 0.332,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6.1 Sol (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-sol",
       "model_display": "GPT-5.6 Sol",
       "priority": 0.477,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.454,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 37.27,
          "norm": 0.474,
          "norm_before_discount": 0.474,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT-5.6 Sol (max)",
          "harness": "Codex",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-sol-max-codex.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.11795450433539938,
          "norm": 0.798,
          "norm_before_discount": 0.798,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Sol (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.0072627731198375844,
          "norm": 0.09,
          "norm_before_discount": 0.09,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Sol (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.6",
       "model_display": "Grok 4.6",
       "priority": 0.45,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.4,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 20.3,
          "norm": 0.15,
          "norm_before_discount": 0.15,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.6 (none)",
          "harness": "Grok Build",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-xai-grok-4-6-none-grok-build.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.09285693951676743,
          "norm": 0.644,
          "norm_before_discount": 0.644,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.6 (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.05680396400430032,
          "norm": 0.405,
          "norm_before_discount": 0.405,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.6 (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-luna",
       "model_display": "GPT-6 Luna",
       "priority": 0.449,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.398,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 16.36,
          "norm": 0.075,
          "norm_before_discount": 0.075,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT-6 Luna (max)",
          "harness": "Codex",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-30-openai-gpt-6-luna-max-codex.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.1090005748897582,
          "norm": 0.743,
          "norm_before_discount": 0.743,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Luna (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.0520701982384284,
          "norm": 0.375,
          "norm_before_discount": 0.375,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Luna (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "gemini-3.8-flash",
       "model_display": "Gemini 3.8 Flash",
       "priority": 0.399,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.298,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 19.09,
          "norm": 0.127,
          "norm_before_discount": 0.127,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Gemini 3.8 Flash (high)",
          "harness": "mini-SWE-agent",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-09-01-gemini-gemini-3-8-flash-high-mini-swe-agent.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.11305488067664629,
          "norm": 0.768,
          "norm_before_discount": 0.768,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Gemini 3.8 Flash (High)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": -0.006867809731099461,
          "norm": 0.0,
          "norm_before_discount": 0.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Gemini 3.8 Flash (High)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-luna",
       "model_display": "GPT-5.6 Luna",
       "priority": 0.366,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.232,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 17.27,
          "norm": 0.093,
          "norm_before_discount": 0.093,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT-5.6 Luna (max)",
          "harness": "Codex",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-luna-max-codex.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.07361256389862167,
          "norm": 0.526,
          "norm_before_discount": 0.526,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Luna (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.005144852705668113,
          "norm": 0.076,
          "norm_before_discount": 0.076,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Luna (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-terra",
       "model_display": "GPT-5.6 Terra",
       "priority": 0.3425,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.185,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 21.52,
          "norm": 0.174,
          "norm_before_discount": 0.174,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT-5.6 Terra (max)",
          "harness": "Codex",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-openai-gpt-5-6-terra-max-codex.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": 0.033616528395016165,
          "norm": 0.28,
          "norm_before_discount": 0.28,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Terra (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.008842784639766732,
          "norm": 0.1,
          "norm_before_discount": 0.1,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Terra (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.5",
       "model_display": "Grok 4.5",
       "priority": 0.3115,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.123,
        "public_default": false,
        "public_metrics": {
         "terminal_bench_4_accuracy": {
          "score": 12.42,
          "norm": 0.0,
          "norm_before_discount": 0.0,
          "harness_relation": "own",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.5 (none)",
          "harness": "Grok Build",
          "source_url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/submissions/2026-08-26-xai-grok-4-5-none-grok-build.json"
         },
         "arena_agent_bash_reliability_no_prereq_overall": {
          "score": -0.011996236994021164,
          "norm": 0.0,
          "norm_before_discount": 0.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.5",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_reliability_no_prereq"
         },
         "arena_agent_bash_recovery_steps_overall": {
          "score": 0.05136066786226423,
          "norm": 0.37,
          "norm_before_discount": 0.37,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.5",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_bash_recovery_steps"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      }
     ]
    },
    "tools": {
     "description": "multi-step work through APIs and tool connectors",
     "metrics_used": [
      "arena_agent_tool_hallucination_overall",
      "arena_agent_task_outcome_explicit_overall"
     ],
     "first_choice": {
      "engine": "codex",
      "engine_name": "Codex CLI",
      "model": "gpt-6.1-sol",
      "model_display": "GPT-6.1 Sol",
      "priority": 0.7145,
      "status": [
       "unproven"
      ],
      "rank_rule": null,
      "tie_broken": false,
      "inputs": {
       "public_norm": 0.929,
       "public_default": false,
       "public_metrics": {
        "arena_agent_tool_hallucination_overall": {
         "score": 0.006459869508645966,
         "norm": 1.0,
         "norm_before_discount": 1.0,
         "harness_relation": "model-only",
         "harness_factor": 1.0,
         "leaderboard_name": "GPT 6.1 Sol (Max)",
         "harness": "",
         "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
        },
        "arena_agent_task_outcome_explicit_overall": {
         "score": 0.1692804702542866,
         "norm": 0.859,
         "norm_before_discount": 0.859,
         "harness_relation": "model-only",
         "harness_factor": 1.0,
         "leaderboard_name": "GPT 6.1 Sol (Max)",
         "harness": "",
         "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
        }
       },
       "own_success": 0.5,
       "own_jobs": 0,
       "own_success_rate": null,
       "own_success_default": true,
       "own_median_minutes": null,
       "speed_rank": 0.5,
       "speed_source": "default"
      }
     },
     "fallbacks": [
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "claude-opus-5.5",
       "model_display": "Claude Opus 5.5",
       "priority": 0.657,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.814,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.005118882708117667,
          "norm": 0.628,
          "norm_before_discount": 0.628,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Opus 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.20481413602090315,
          "norm": 1.0,
          "norm_before_discount": 1.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Opus 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-opus-5.5",
       "model_display": "Claude Opus 5.5",
       "priority": 0.657,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.814,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.005118882708117667,
          "norm": 0.628,
          "norm_before_discount": 0.628,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Opus 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.20481413602090315,
          "norm": 1.0,
          "norm_before_discount": 1.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Opus 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-fable-5.1",
       "model_display": "Claude Fable 5.1",
       "priority": 0.649,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.798,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.005647594164088332,
          "norm": 0.775,
          "norm_before_discount": 0.775,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Fable 5.1 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.15951173978047029,
          "norm": 0.821,
          "norm_before_discount": 0.821,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Fable 5.1 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-astra",
       "model_display": "GPT-6 Astra",
       "priority": 0.611,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.722,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.00583104928128364,
          "norm": 0.826,
          "norm_before_discount": 0.826,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Astra (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.10836770521757655,
          "norm": 0.619,
          "norm_before_discount": 0.619,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Astra (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "claude-sonnet-5.5",
       "model_display": "Claude Sonnet 5.5",
       "priority": 0.5885,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.677,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.00438080270551211,
          "norm": 0.424,
          "norm_before_discount": 0.424,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Sonnet 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.18699556464585243,
          "norm": 0.93,
          "norm_before_discount": 0.93,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Sonnet 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "claude",
       "engine_name": "Claude Code",
       "model": "claude-sonnet-5.5",
       "model_display": "Claude Sonnet 5.5",
       "priority": 0.5885,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.677,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.00438080270551211,
          "norm": 0.424,
          "norm_before_discount": 0.424,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Sonnet 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.18699556464585243,
          "norm": 0.93,
          "norm_before_discount": 0.93,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Claude Sonnet 5.5 (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.7",
       "model_display": "Grok 4.7",
       "priority": 0.537,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.574,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.005500392769675955,
          "norm": 0.734,
          "norm_before_discount": 0.734,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.7 (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.05647377469876602,
          "norm": 0.413,
          "norm_before_discount": 0.413,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.7 (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.5",
       "model_display": "Grok 4.5",
       "priority": 0.5155,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.531,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.005886244468655377,
          "norm": 0.841,
          "norm_before_discount": 0.841,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.5",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.007945250315153807,
          "norm": 0.221,
          "norm_before_discount": 0.221,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.5",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "agy",
       "engine_name": "Antigravity CLI",
       "model": "gemini-3.8-flash",
       "model_display": "Gemini 3.8 Flash",
       "priority": 0.508,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.516,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.0050338628294710475,
          "norm": 0.605,
          "norm_before_discount": 0.605,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Gemini 3.8 Flash (High)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.06016740039705126,
          "norm": 0.428,
          "norm_before_discount": 0.428,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Gemini 3.8 Flash (High)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-sol",
       "model_display": "GPT-5.6 Sol",
       "priority": 0.4995,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.499,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.005036197285249955,
          "norm": 0.605,
          "norm_before_discount": 0.605,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Sol (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.05132260089878238,
          "norm": 0.393,
          "norm_before_discount": 0.393,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Sol (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "grok",
       "engine_name": "Grok CLI",
       "model": "grok-4.6",
       "model_display": "Grok 4.6",
       "priority": 0.4525,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.405,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.005137515593574449,
          "norm": 0.633,
          "norm_before_discount": 0.633,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.6 (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": -0.0030479395306920454,
          "norm": 0.178,
          "norm_before_discount": 0.178,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "Grok 4.6 (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-luna",
       "model_display": "GPT-6 Luna",
       "priority": 0.382,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.264,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.003895247049200634,
          "norm": 0.289,
          "norm_before_discount": 0.289,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Luna (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.012743296730492237,
          "norm": 0.24,
          "norm_before_discount": 0.24,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Luna (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-6-sol",
       "model_display": "GPT-6 Sol",
       "priority": 0.3735,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.247,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.0028519862142601937,
          "norm": 0.0,
          "norm_before_discount": 0.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Sol (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": 0.07697932975685551,
          "norm": 0.494,
          "norm_before_discount": 0.494,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 6 Sol (Max)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-luna",
       "model_display": "GPT-5.6 Luna",
       "priority": 0.348,
       "status": [
        "unproven"
       ],
       "rank_rule": "tie",
       "tie_broken": true,
       "inputs": {
        "public_norm": 0.196,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.0042632553447887744,
          "norm": 0.391,
          "norm_before_discount": 0.391,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Luna (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": -0.048055579182375796,
          "norm": 0.0,
          "norm_before_discount": 0.0,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Luna (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      },
      {
       "engine": "codex",
       "engine_name": "Codex CLI",
       "model": "gpt-5.6-terra",
       "model_display": "GPT-5.6 Terra",
       "priority": 0.3255,
       "status": [
        "unproven"
       ],
       "rank_rule": null,
       "tie_broken": false,
       "inputs": {
        "public_norm": 0.151,
        "public_default": false,
        "public_metrics": {
         "arena_agent_tool_hallucination_overall": {
          "score": 0.0038333798804330854,
          "norm": 0.272,
          "norm_before_discount": 0.272,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Terra (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_tool_hallucination"
         },
         "arena_agent_task_outcome_explicit_overall": {
          "score": -0.04080794119134068,
          "norm": 0.029,
          "norm_before_discount": 0.029,
          "harness_relation": "model-only",
          "harness_factor": 1.0,
          "leaderboard_name": "GPT 5.6 Terra (xHigh)",
          "harness": "",
          "source_url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/viewer/agent_task_outcome_explicit"
         }
        },
        "own_success": 0.5,
        "own_jobs": 0,
        "own_success_rate": null,
        "own_success_default": true,
        "own_median_minutes": null,
        "speed_rank": 0.5,
        "speed_source": "default"
       }
      }
     ]
    }
   }
  },
  "research": {
   "description": "web research and fact-finding",
   "metrics_used": [
    "drb_total_score*",
    "arena_search_overall"
   ],
   "first_choice": {
    "engine": "codex",
    "engine_name": "Codex CLI",
    "model": "gpt-5.6-sol",
    "model_display": "GPT-5.6 Sol",
    "priority": 0.75,
    "status": [
     "unproven"
    ],
    "rank_rule": null,
    "tie_broken": false,
    "inputs": {
     "public_norm": 1.0,
     "public_default": false,
     "public_metrics": {
      "arena_search_overall": {
       "score": 1257.2603500980206,
       "norm": 1.0,
       "norm_before_discount": 1.0,
       "harness_relation": "model-only",
       "harness_factor": 1.0,
       "leaderboard_name": "gpt-5.6-sol-xhigh",
       "harness": "",
       "source_url": "https://arena.ai/leaderboard/search"
      }
     },
     "own_success": 0.5,
     "own_jobs": 0,
     "own_success_rate": null,
     "own_success_default": true,
     "own_median_minutes": null,
     "speed_rank": 0.5,
     "speed_source": "default"
    }
   },
   "fallbacks": [
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-sonnet-5.5",
     "model_display": "Claude Sonnet 5.5",
     "priority": 0.6185,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.895,
      "own_jobs": 19,
      "own_success_rate": 0.895,
      "own_success_default": false,
      "own_median_minutes": 5.0,
      "speed_rank": 0.5,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-haiku-5.5",
     "model_display": "Claude Haiku 5.5",
     "priority": 0.6,
     "status": [
      "proven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.5,
      "own_jobs": 2,
      "own_success_rate": 1.0,
      "own_success_default": true,
      "own_median_minutes": 0.4,
      "speed_rank": 1.0,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-fable-5.1",
     "model_display": "Claude Fable 5.1",
     "priority": 0.55,
     "status": [
      "proven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.5,
      "own_jobs": 7,
      "own_success_rate": 0.571,
      "own_success_default": true,
      "own_median_minutes": 1.4,
      "speed_rank": 0.75,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6.1-sol",
     "model_display": "GPT-6.1 Sol",
     "priority": 0.4213,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.571,
      "own_jobs": 28,
      "own_success_rate": 0.571,
      "own_success_default": false,
      "own_median_minutes": 13.7,
      "speed_rank": 0.0,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-sol",
     "model_display": "GPT-6 Sol",
     "priority": 0.45,
     "status": [
      "proven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.5,
      "own_jobs": 1,
      "own_success_rate": 1.0,
      "own_success_default": true,
      "own_median_minutes": 6.9,
      "speed_rank": 0.25,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.5",
     "model_display": "Grok 4.5",
     "priority": 0.25,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.0,
      "public_default": false,
      "public_metrics": {
       "arena_search_overall": {
        "score": 1212.646185108397,
        "norm": 0.0,
        "norm_before_discount": 0.0,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.5",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/search"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    }
   ]
  },
  "reading": {
   "description": "long documents, transcripts, summaries, extraction",
   "metrics_used": [
    "arena_document_overall"
   ],
   "first_choice": {
    "engine": "claude",
    "engine_name": "Claude Code",
    "model": "claude-fable-5.1",
    "model_display": "Claude Fable 5.1",
    "priority": 0.75,
    "status": [
     "unproven"
    ],
    "rank_rule": null,
    "tie_broken": false,
    "inputs": {
     "public_norm": 1.0,
     "public_default": false,
     "public_metrics": {
      "arena_document_overall": {
       "score": 1512.5843185623094,
       "norm": 1.0,
       "norm_before_discount": 1.0,
       "harness_relation": "model-only",
       "harness_factor": 1.0,
       "leaderboard_name": "claude-fable-5.1-max",
       "harness": "",
       "source_url": "https://arena.ai/leaderboard/document"
      }
     },
     "own_success": 0.5,
     "own_jobs": 0,
     "own_success_rate": null,
     "own_success_default": true,
     "own_median_minutes": null,
     "speed_rank": 0.5,
     "speed_source": "default"
    }
   },
   "fallbacks": [
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-sonnet-5.5",
     "model_display": "Claude Sonnet 5.5",
     "priority": 0.6,
     "status": [
      "proven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.5,
      "own_jobs": 5,
      "own_success_rate": 1.0,
      "own_success_default": true,
      "own_median_minutes": 1.3,
      "speed_rank": 1.0,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6.1-sol",
     "model_display": "GPT-6.1 Sol",
     "priority": 0.49,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.8,
      "own_jobs": 15,
      "own_success_rate": 0.8,
      "own_success_default": false,
      "own_median_minutes": 9.1,
      "speed_rank": 0.0,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-sol",
     "model_display": "GPT-5.6 Sol",
     "priority": 0.5035,
     "status": [
      "proven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.507,
      "public_default": false,
      "public_metrics": {
       "arena_document_overall": {
        "score": 1482.7123255544448,
        "norm": 0.507,
        "norm_before_discount": 0.507,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-sol-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/document"
       }
      },
      "own_success": 0.5,
      "own_jobs": 2,
      "own_success_rate": 1.0,
      "own_success_default": true,
      "own_median_minutes": 5.8,
      "speed_rank": 0.5,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-terra",
     "model_display": "GPT-5.6 Terra",
     "priority": 0.412,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.324,
      "public_default": false,
      "public_metrics": {
       "arena_document_overall": {
        "score": 1471.5811447461108,
        "norm": 0.324,
        "norm_before_discount": 0.324,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-terra-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/document"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-astra",
     "model_display": "GPT-6 Astra",
     "priority": 0.3785,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.257,
      "public_default": false,
      "public_metrics": {
       "arena_document_overall": {
        "score": 1467.5231106903996,
        "norm": 0.257,
        "norm_before_discount": 0.257,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6-astra-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/document"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-luna",
     "model_display": "GPT-5.6 Luna",
     "priority": 0.291,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.082,
      "public_default": false,
      "public_metrics": {
       "arena_document_overall": {
        "score": 1456.916653229613,
        "norm": 0.082,
        "norm_before_discount": 0.082,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-luna-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/document"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.6",
     "model_display": "Grok 4.6",
     "priority": 0.253,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.006,
      "public_default": false,
      "public_metrics": {
       "arena_document_overall": {
        "score": 1452.3297871123095,
        "norm": 0.006,
        "norm_before_discount": 0.006,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.6-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/document"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.5",
     "model_display": "Grok 4.5",
     "priority": 0.25,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.0,
      "public_default": false,
      "public_metrics": {
       "arena_document_overall": {
        "score": 1451.9590443080492,
        "norm": 0.0,
        "norm_before_discount": 0.0,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.5",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/document"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    }
   ]
  },
  "writing": {
   "description": "drafting and editing prose",
   "metrics_used": [
    "arena_text_creative_writing",
    "arena_text_overall"
   ],
   "first_choice": {
    "engine": "claude",
    "engine_name": "Claude Code",
    "model": "claude-sonnet-5.5",
    "model_display": "Claude Sonnet 5.5",
    "priority": 0.6444,
    "status": [
     "proven"
    ],
    "rank_rule": "evidence",
    "tie_broken": false,
    "inputs": {
     "public_norm": 0.722,
     "public_default": false,
     "public_metrics": {
      "arena_text_overall": {
       "score": 1471.1563597911008,
       "norm": 0.708,
       "norm_before_discount": 0.708,
       "harness_relation": "model-only",
       "harness_factor": 1.0,
       "leaderboard_name": "claude-sonnet-5.5-xhigh",
       "harness": "",
       "source_url": "https://arena.ai/leaderboard/text"
      },
      "arena_text_creative_writing": {
       "score": 1464.5946985847822,
       "norm": 0.736,
       "norm_before_discount": 0.736,
       "harness_relation": "model-only",
       "harness_factor": 1.0,
       "leaderboard_name": "claude-sonnet-5.5-xhigh",
       "harness": "",
       "source_url": "https://arena.ai/leaderboard/text"
      }
     },
     "own_success": 0.5,
     "own_jobs": 2,
     "own_success_rate": 1.0,
     "own_success_default": true,
     "own_median_minutes": 1.3,
     "speed_rank": 0.667,
     "speed_source": "own median minutes"
    }
   },
   "fallbacks": [
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-opus-5.5",
     "model_display": "Claude Opus 5.5",
     "priority": 0.65,
     "status": [
      "proven"
     ],
     "rank_rule": "evidence",
     "tie_broken": false,
     "inputs": {
      "public_norm": 1.0,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1514.8395928693265,
        "norm": 1.0,
        "norm_before_discount": 1.0,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "claude-opus-5.5-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1532.7119224706557,
        "norm": 1.0,
        "norm_before_discount": 1.0,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "claude-opus-5.5-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 2,
      "own_success_rate": 1.0,
      "own_success_default": true,
      "own_median_minutes": 2.4,
      "speed_rank": 0.0,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "claude-opus-5.5",
     "model_display": "Claude Opus 5.5",
     "priority": 0.75,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 1.0,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1514.8395928693265,
        "norm": 1.0,
        "norm_before_discount": 1.0,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "claude-opus-5.5-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1532.7119224706557,
        "norm": 1.0,
        "norm_before_discount": 1.0,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "claude-opus-5.5-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-fable-5.1",
     "model_display": "Claude Fable 5.1",
     "priority": 0.7165,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.933,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1510.0200778498338,
        "norm": 0.968,
        "norm_before_discount": 0.968,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "claude-fable-5.1-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1506.5205687318962,
        "norm": 0.899,
        "norm_before_discount": 0.899,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "claude-fable-5.1-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "gemini-3.8-flash",
     "model_display": "Gemini 3.8 Flash",
     "priority": 0.684,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.868,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1498.9549657523576,
        "norm": 0.894,
        "norm_before_discount": 0.894,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gemini-3.8-flash-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1491.9444570095316,
        "norm": 0.842,
        "norm_before_discount": 0.842,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gemini-3.8-flash-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "claude",
     "engine_name": "Claude Code",
     "model": "claude-haiku-5.5",
     "model_display": "Claude Haiku 5.5",
     "priority": 0.6,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.5,
      "public_default": true,
      "public_metrics": {},
      "own_success": 0.5,
      "own_jobs": 1,
      "own_success_rate": 1.0,
      "own_success_default": true,
      "own_median_minutes": 0.0,
      "speed_rank": 1.0,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "claude-sonnet-5.5",
     "model_display": "Claude Sonnet 5.5",
     "priority": 0.611,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.722,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1471.1563597911008,
        "norm": 0.708,
        "norm_before_discount": 0.708,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "claude-sonnet-5.5-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1464.5946985847822,
        "norm": 0.736,
        "norm_before_discount": 0.736,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "claude-sonnet-5.5-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-sol",
     "model_display": "GPT-5.6 Sol",
     "priority": 0.5715,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.643,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1457.3398514233177,
        "norm": 0.615,
        "norm_before_discount": 0.615,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-sol-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1447.8249887716902,
        "norm": 0.671,
        "norm_before_discount": 0.671,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-sol-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.5",
     "model_display": "Grok 4.5",
     "priority": 0.5505,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.601,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1448.1066515962236,
        "norm": 0.553,
        "norm_before_discount": 0.553,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.5",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1442.0606065511893,
        "norm": 0.649,
        "norm_before_discount": 0.649,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.5",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6.1-sol",
     "model_display": "GPT-6.1 Sol",
     "priority": 0.5051,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.577,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1446.50994938648,
        "norm": 0.543,
        "norm_before_discount": 0.543,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6.1-sol-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1432.1009071720614,
        "norm": 0.61,
        "norm_before_discount": 0.61,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6.1-sol-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 6,
      "own_success_rate": 0.833,
      "own_success_default": true,
      "own_median_minutes": 2.2,
      "speed_rank": 0.333,
      "speed_source": "own median minutes"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-terra",
     "model_display": "GPT-5.6 Terra",
     "priority": 0.517,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.534,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1446.536232387477,
        "norm": 0.543,
        "norm_before_discount": 0.543,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-terra-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1409.7882877909296,
        "norm": 0.524,
        "norm_before_discount": 0.524,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-terra-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-astra",
     "model_display": "GPT-6 Astra",
     "priority": 0.515,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.53,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1440.473402974347,
        "norm": 0.502,
        "norm_before_discount": 0.502,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6-astra-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1418.2456422308837,
        "norm": 0.557,
        "norm_before_discount": 0.557,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6-astra-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.6",
     "model_display": "Grok 4.6",
     "priority": 0.503,
     "status": [
      "unproven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.506,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1427.8462039982326,
        "norm": 0.418,
        "norm_before_discount": 0.418,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.6-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1427.841125927157,
        "norm": 0.594,
        "norm_before_discount": 0.594,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.6-high",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-5.6-luna",
     "model_display": "GPT-5.6 Luna",
     "priority": 0.4765,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.453,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1430.4567491378307,
        "norm": 0.435,
        "norm_before_discount": 0.435,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-luna-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1395.7940087483485,
        "norm": 0.47,
        "norm_before_discount": 0.47,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-5.6-luna-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-sol",
     "model_display": "GPT-6 Sol",
     "priority": 0.3995,
     "status": [
      "proven"
     ],
     "rank_rule": "tie",
     "tie_broken": true,
     "inputs": {
      "public_norm": 0.299,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1394.9679705033293,
        "norm": 0.198,
        "norm_before_discount": 0.198,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6-sol-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1377.9475717899409,
        "norm": 0.401,
        "norm_before_discount": 0.401,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6-sol-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 1,
      "own_success_rate": 0.0,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "grok",
     "engine_name": "Grok CLI",
     "model": "grok-4.7",
     "model_display": "Grok 4.7",
     "priority": 0.4195,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.339,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1399.344695615349,
        "norm": 0.227,
        "norm_before_discount": 0.227,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.7-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1390.921572688785,
        "norm": 0.451,
        "norm_before_discount": 0.451,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "grok-4.7-xhigh",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "codex",
     "engine_name": "Codex CLI",
     "model": "gpt-6-luna",
     "model_display": "GPT-6 Luna",
     "priority": 0.3785,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.257,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1391.0651807781835,
        "norm": 0.172,
        "norm_before_discount": 0.172,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6-luna-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1362.757551552419,
        "norm": 0.342,
        "norm_before_discount": 0.342,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-6-luna-max",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    },
    {
     "engine": "agy",
     "engine_name": "Antigravity CLI",
     "model": "gpt-oss-120b",
     "model_display": "GPT-OSS 120B",
     "priority": 0.25,
     "status": [
      "unproven"
     ],
     "rank_rule": null,
     "tie_broken": false,
     "inputs": {
      "public_norm": 0.0,
      "public_default": false,
      "public_metrics": {
       "arena_text_overall": {
        "score": 1365.421841082085,
        "norm": 0.0,
        "norm_before_discount": 0.0,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-oss-120b",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       },
       "arena_text_creative_writing": {
        "score": 1274.5045587868208,
        "norm": 0.0,
        "norm_before_discount": 0.0,
        "harness_relation": "model-only",
        "harness_factor": 1.0,
        "leaderboard_name": "gpt-oss-120b",
        "harness": "",
        "source_url": "https://arena.ai/leaderboard/text"
       }
      },
      "own_success": 0.5,
      "own_jobs": 0,
      "own_success_rate": null,
      "own_success_default": true,
      "own_median_minutes": null,
      "speed_rank": 0.5,
      "speed_source": "default"
     }
    }
   ]
  }
 },
 "sources": [
  {
   "title": "Terminal-Bench 4.0",
   "url": "https://github.com/harbor-framework/terminal-bench/blob/main/leaderboard/leaderboard.yaml"
  },
  {
   "title": "SWE-bench",
   "url": "https://www.swebench.com/"
  },
  {
   "title": "DeepResearch Bench",
   "url": "https://huggingface.co/spaces/muset-ai/DeepResearch-Bench-Leaderboard"
  },
  {
   "title": "Arena (text, document, search, webdev, agent)",
   "url": "https://arena.ai/leaderboard"
  }
 ]
}