{
  "schema": 1,
  "exp_id": "exp-20260915-ep1-f005-docker",
  "title": "EP1 docker 冻结夹具受控实验（ep1-f005 × 9 模型：flash 3 + 御三家旗舰 2 + 多厂旗舰 4；+ p1 冒烟）",
  "generated_at": "2026-09-21T08:54:26+00:00",
  "generator": "scripts/snapshot-experiment.py（重跑可再生；禁止手改本文件）",
  "design": {
    "independent_variable": "model",
    "note": "控制变量硬要求见 schema/SCHEMA.md「实验环境与隔离」节；夹具见 fixture 块"
  },
  "fixture": {
    "schema": 1,
    "frozen_at": "2026-09-15T06:41:51+00:00",
    "image_tag": "0genlab/hermes-harness:ep1",
    "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
    "base_image": "unknown",
    "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
    "hermes_version_in_image": "Hermes Agent v0.20.5 (2026.8.19)",
    "python_in_image": "3.11.16",
    "anthropic_in_image": "0.94.0",
    "node_in_image": null,
    "adapter_sha256": "adf39193da22331a4e6cff16b5470c012cf1d23ce43d32a625794d999fd28334",
    "leak_guard": "pass: adapter 无 F-005 修复模式",
    "limits": {
      "cpus": "2",
      "memory": "4g",
      "pids_limit": 512
    },
    "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
  },
  "runs": [
    {
      "run_id": "hermes-deepseekv4.1flash-ep1-f005-r7",
      "model": "deepseek-v4.1-flash",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 7,
      "score": 1,
      "wall_s": 242,
      "agent_exit": 0,
      "ts": "2026-09-15T06:49:17+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "ca7028c8ff8df9fb77a275223f9ca9dc1d8dbbcf1f5ede8456643870b4edc924",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 20,
        "input_tokens": 53437,
        "output_tokens": 8986,
        "cache_read_tokens": 678784,
        "total_tokens": 741207
      }
    },
    {
      "run_id": "hermes-deepseekv4.1flash-ep1-f005-r8",
      "model": "deepseek-v4.1-flash",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 8,
      "score": 1,
      "wall_s": 196,
      "agent_exit": 0,
      "ts": "2026-09-15T09:06:14+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "7d4f6d6bec80be09302bd915812be3a6cd4cf32abef0d7ce2f2afe109b948791",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 16,
        "input_tokens": 46697,
        "output_tokens": 4615,
        "cache_read_tokens": 449408,
        "total_tokens": 500720
      }
    },
    {
      "run_id": "hermes-deepseekv4pro-ep1-f005-r11",
      "model": "deepseek-v4-pro",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 11,
      "score": 1,
      "wall_s": 239,
      "agent_exit": 0,
      "ts": "2026-09-17T15:02:19+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "53221ef883266d2766def120d56fa055b13ce807c17a8bbe556d66a4bd9536aa",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 12,
        "input_tokens": 160892,
        "output_tokens": 4948,
        "cache_read_tokens": 194816,
        "total_tokens": 360656
      }
    },
    {
      "run_id": "hermes-deepseekv4pro-ep1-f005-r12",
      "model": "deepseek-v4-pro",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 12,
      "score": 1,
      "wall_s": 381,
      "agent_exit": 0,
      "ts": "2026-09-17T15:09:05+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "284dc17a2bd67efacd37563a698065fbb54255a938f175c4694f92ba77623a7d",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 24,
        "input_tokens": 221785,
        "output_tokens": 6183,
        "cache_read_tokens": 503040,
        "total_tokens": 731008
      }
    },
    {
      "run_id": "hermes-gemini3.8flash-ep1-f005-r7",
      "model": "gemini-3.8-flash",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 7,
      "score": 1,
      "wall_s": 187,
      "agent_exit": 0,
      "ts": "2026-09-15T06:53:15+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "5726be3876cb6537e1717ae96749b5032fb1d4a5609add7a4dc5e244b7070f18",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 26,
        "input_tokens": 169576,
        "output_tokens": 2499,
        "cache_read_tokens": 566958,
        "total_tokens": 739033
      }
    },
    {
      "run_id": "hermes-gemini3.8flash-ep1-f005-r8",
      "model": "gemini-3.8-flash",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 8,
      "score": 1,
      "wall_s": 269,
      "agent_exit": 0,
      "ts": "2026-09-15T09:11:09+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "41dd8f63337c620705ce3461968a3a741e258a8cb5fe57d8aa6fcc9882ea7e78",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 25,
        "input_tokens": 277685,
        "output_tokens": 3850,
        "cache_read_tokens": 309843,
        "total_tokens": 591378
      }
    },
    {
      "run_id": "hermes-glm5.3flash-ep1-f005-r7",
      "model": "glm-5.3-flash",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 7,
      "score": 0,
      "wall_s": 426,
      "agent_exit": 124,
      "ts": "2026-09-15T07:01:00+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "251e43b6f700aec8187fb2352c1aae12aa65ef8d435cecdcb428ccfc0e8e0046",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": false,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": null,
        "input_tokens": null,
        "output_tokens": null,
        "cache_read_tokens": null,
        "total_tokens": null
      },
      "wire_backfill": {
        "run": "hermes-glm5.3flash-ep1-f005-r7",
        "api_calls": 14,
        "net_input": 49899,
        "output": 14392,
        "cache_read": 266560,
        "cache_write": 0,
        "gross_input": 316459,
        "reasoning": 12623,
        "cache_rate_pct": 84,
        "_note": "exit124 未写官方账本，此块由 relay 录制重算（scripts/agg-usage.mjs）"
      }
    },
    {
      "run_id": "hermes-glm5.3flash-ep1-f005-r8",
      "model": "glm-5.3-flash",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 8,
      "score": 0,
      "wall_s": 425,
      "agent_exit": 124,
      "ts": "2026-09-15T09:18:38+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "cabd52ca343e72c99ca7211a9eac436fb7299a94e476e2908fc40f3d2b300dbe",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": false,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": null,
        "input_tokens": null,
        "output_tokens": null,
        "cache_read_tokens": null,
        "total_tokens": null
      },
      "wire_backfill": {
        "run": "hermes-glm5.3flash-ep1-f005-r8",
        "api_calls": 9,
        "net_input": 26283,
        "output": 13761,
        "cache_read": 138752,
        "cache_write": 0,
        "gross_input": 165035,
        "reasoning": 12496,
        "cache_rate_pct": 84,
        "_note": "exit124 未写官方账本，此块由 relay 录制重算（scripts/agg-usage.mjs）"
      }
    },
    {
      "run_id": "hermes-glm5.3-ep1-f005-r13",
      "model": "glm-5.3",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 13,
      "score": 0,
      "wall_s": 423,
      "agent_exit": 124,
      "ts": "2026-09-17T15:16:46+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "5ab430a663c871272b6f769af545d072bd6bba430c2ca8617fca00ca67805b35",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": false,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": null,
        "input_tokens": null,
        "output_tokens": null,
        "cache_read_tokens": null,
        "total_tokens": null
      },
      "wire_backfill": {
        "run": "hermes-glm5.3-ep1-f005-r13",
        "api_calls": 25,
        "net_input": 49672,
        "output": 32072,
        "cache_read": 523264,
        "cache_write": 0,
        "gross_input": 572936,
        "reasoning": 30261,
        "cache_rate_pct": 91,
        "_note": "exit124 未写官方账本，此块由 relay 录制重算（scripts/agg-usage.mjs）"
      }
    },
    {
      "run_id": "hermes-glm5.3-ep1-f005-r14",
      "model": "glm-5.3",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 14,
      "score": 0,
      "wall_s": 424,
      "agent_exit": 124,
      "ts": "2026-09-17T15:24:29+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "dbf428b90c3eff342b76e4d4f42b57eb35b9e4b4379339a0c58b4fa5da9f1bee",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": false,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": null,
        "input_tokens": null,
        "output_tokens": null,
        "cache_read_tokens": null,
        "total_tokens": null
      },
      "wire_backfill": {
        "run": "hermes-glm5.3-ep1-f005-r14",
        "api_calls": 25,
        "net_input": 22973,
        "output": 36123,
        "cache_read": 528640,
        "cache_write": 0,
        "gross_input": 551613,
        "reasoning": 33918,
        "cache_rate_pct": 96,
        "_note": "exit124 未写官方账本，此块由 relay 录制重算（scripts/agg-usage.mjs）"
      }
    },
    {
      "run_id": "hermes-claudefable51-ep1-f005-r9",
      "model": "claude-fable-5-1",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 9,
      "score": 1,
      "wall_s": 142,
      "agent_exit": 0,
      "ts": "2026-09-17T14:36:32+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "74db09f82e33ca0d754ab239b3135f75ff1b86e0be2b2730b2dd8ac9dba2cb6c",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 9,
        "input_tokens": 209021,
        "output_tokens": 2864,
        "cache_read_tokens": 0,
        "total_tokens": 211885
      }
    },
    {
      "run_id": "hermes-claudefable51-ep1-f005-r10",
      "model": "claude-fable-5-1",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 10,
      "score": 1,
      "wall_s": 140,
      "agent_exit": 0,
      "ts": "2026-09-17T14:39:23+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "8deaedd7ef59465215493bc8348a52713fd3dfb83f161df632a1ec968ad6b260",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 10,
        "input_tokens": 238926,
        "output_tokens": 2916,
        "cache_read_tokens": 0,
        "total_tokens": 241842
      }
    },
    {
      "run_id": "hermes-gpt6astra-ep1-f005-r9",
      "model": "gpt-6-astra",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 9,
      "score": 1,
      "wall_s": 83,
      "agent_exit": 0,
      "ts": "2026-09-17T14:41:10+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "19c1c9fb76980a4608bdc8a38a2d38a8231070c29e9dbc28762a36ad21ccf039",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 8,
        "input_tokens": 24,
        "output_tokens": 469,
        "cache_read_tokens": 98111,
        "total_tokens": 120942
      }
    },
    {
      "run_id": "hermes-gpt6astra-ep1-f005-r10",
      "model": "gpt-6-astra",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 10,
      "score": 1,
      "wall_s": 86,
      "agent_exit": 0,
      "ts": "2026-09-17T14:43:07+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "f06ab75a766ef07b0f8ae33256e0ba43bc1de4fe82bcae8b27d2de068bc6bf16",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 8,
        "input_tokens": 24,
        "output_tokens": 474,
        "cache_read_tokens": 108866,
        "total_tokens": 120993
      }
    },
    {
      "run_id": "hermes-kimik3-ep1-f005-r15",
      "model": "kimi-k3",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 15,
      "score": 1,
      "wall_s": 383,
      "agent_exit": 0,
      "ts": "2026-09-17T15:31:26+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "e568bce342453d270e9fe49ae805c2271d43c8d9255535f8a63be2eba2185ed5",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 13,
        "input_tokens": 99312,
        "output_tokens": 6361,
        "cache_read_tokens": 103944,
        "total_tokens": 209617
      }
    },
    {
      "run_id": "hermes-kimik3-ep1-f005-r16",
      "model": "kimi-k3",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 16,
      "score": 1,
      "wall_s": 199,
      "agent_exit": 0,
      "ts": "2026-09-17T15:35:14+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "f6e37c683c6f04d65a7359124e224a35e0f38e99f7ebb879073c9d31037bc896",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 15,
        "input_tokens": 65396,
        "output_tokens": 2625,
        "cache_read_tokens": 183010,
        "total_tokens": 251031
      }
    },
    {
      "run_id": "hermes-grok4.6-ep1-f005-r17",
      "model": "grok-4.6",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 17,
      "score": 1,
      "wall_s": 144,
      "agent_exit": 0,
      "ts": "2026-09-17T15:38:09+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "74f10e5292da28db3ae7ad1a05ff9c40f0eb411721dea0931df83f7e288315ef",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 14,
        "input_tokens": 52246,
        "output_tokens": 1746,
        "cache_read_tokens": 242944,
        "total_tokens": 296936
      }
    },
    {
      "run_id": "hermes-grok4.6-ep1-f005-r18",
      "model": "grok-4.6",
      "task": "ep1-f005",
      "task_type": "source",
      "attempt": 18,
      "score": 1,
      "wall_s": 119,
      "agent_exit": 0,
      "ts": "2026-09-17T15:40:43+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 420,
        "max_turns": 20,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "72519a2b2c671c9a362a8f0a7e1b2c6934b4b3cbb62ac05288a881a9f4e98cc7",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {
        "test_passed": true,
        "test_modified_by_model": false,
        "py": "python3.11"
      },
      "official_usage": {
        "api_calls": 12,
        "input_tokens": 27546,
        "output_tokens": 1513,
        "cache_read_tokens": 215424,
        "total_tokens": 244483
      }
    },
    {
      "run_id": "hermes-deepseekv4.1flash-p1-r2",
      "model": "deepseek-v4.1-flash",
      "task": "p1",
      "task_type": "answer",
      "attempt": 2,
      "score": 1,
      "wall_s": 102,
      "agent_exit": 0,
      "ts": "2026-09-15T06:44:31+00:00",
      "params": {
        "provider": "aihubmix",
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "harness": "docker",
        "timeout_s": 600,
        "max_turns": 40,
        "sandbox": false,
        "strict_home": false
      },
      "env": {
        "host": {
          "os": "Darwin 25.4.0",
          "arch": "arm64"
        },
        "agent": "Hermes Agent v0.20.5 (2026.8.19)",
        "tool_versions": {
          "python3": "3.11.16",
          "node": null,
          "anthropic": "0.94.0"
        },
        "sandbox_kind": "docker",
        "env_policy": "container-clean: 容器 env 仅显式注入 KEY_ENV+HERMES_HOME，宿主凭据不存在",
        "docker": {
          "image_tag": "0genlab/hermes-harness:ep1",
          "image_id": "sha256:42357c39e76d906d451ff2fb54c88666078691d6e5b4787f5476a0faf00fb003",
          "base_image": "unknown",
          "hermes_commit": "4c1f53be10d0fce1d25aee1975e5149b6c54f25a",
          "container_id": "ca1a2c166cb323b9371f2fc823ca75f0aec952a9da10744fc9585d6e0e3aeb2c",
          "limits": {
            "cpus": "2",
            "memory": "4g",
            "pids_limit": 512
          },
          "network": "0genlab-int (internal; 出网仅经 0genlab-relaygw 边车到宿主 relay:8484)"
        }
      },
      "score_diag": {},
      "official_usage": {
        "api_calls": 8,
        "input_tokens": 16853,
        "output_tokens": 1733,
        "cache_read_tokens": 93312,
        "total_tokens": 111898
      }
    }
  ],
  "notes": [
    "2026-09-17 追加多厂旗舰批（attempt r11-r18，同冻结夹具、模型为唯一自变量）：deepseek-v4-pro 2/2(239/381s,方差大,缓存率高)；glm-5.3 0/2(双 exit124 撞 420s 墙钟)——旗舰与 flash 同样结构性卡死，GLM 不分档位在本任务收敛不了；kimi-k3(月之暗面)2/2(383/199s)；grok-4.6(xAI)2/2(144/119s,快而稳,仅次于 gpt-6-astra)。",
    "同厂跨档对照：deepseek flash(v4.1-flash) 与 旗舰(v4-pro) 均 2/2、旗舰方差更大；glm flash(5.3-flash) 与 旗舰(5.3) 均 0/2 撞墙——档位升级未改变 GLM 的失败模式。",
    "档位/代际不齐须标注：gemini 保持 3.8-flash（网关无 3.8-max/pro，用户拍板不换）；fable-5-1 与 gpt-6-astra 为御三家旗舰；deepseek-v4-pro/glm-5.3/kimi-k3/grok-4.6 为各厂旗舰。横比成本/时延须按档位分组，勿跨档直接排名（校准层红线：不出榜）。"
  ]
}