{
  "name": "nine models, three trials each, bare agent",
  "description": "Every model on the same subject, same harness, three trials each so the spread is measured rather than assumed",
  "defaults": {
    "pr": "smykla-skalski/cairn#8",
    "subject": "proxy",
    "mode": "agent",
    "effort": "high",
    "key_limit": 2
  },
  "runs": [
    {"model": "openrouter:openai/gpt-oss-120b", "label": "gpt-oss-120b", "repeat": 3},
    {"model": "openrouter:qwen/qwen3-coder-30b-a3b-instruct", "label": "qwen3-coder-30b", "repeat": 3},
    {"model": "openrouter:z-ai/glm-5.3-flash", "label": "glm-5.3-flash", "repeat": 3},
    {"model": "openrouter:deepseek/deepseek-v4-flash", "label": "deepseek-v4-flash", "repeat": 3},
    {"model": "openrouter:moonshotai/kimi-k2.7-code", "label": "kimi-k2.7-code", "repeat": 3},
    {"model": "openrouter:z-ai/glm-5.2", "label": "glm-5.2", "repeat": 3},
    {"model": "openrouter:z-ai/glm-5.3", "label": "glm-5.3", "repeat": 3},
    {"model": "openrouter:deepseek/deepseek-v4-pro", "label": "deepseek-v4-pro", "repeat": 3},
    {"model": "openrouter:x-ai/grok-4.5", "label": "grok-4.5", "repeat": 3}
  ]
}
