{
  "model_name": "Gemini 3.0 Pro",
  "model_organization": "Google",
  "submitting_organization": "Sierra",
  "submission_date": "2025-11-18",
  "contact_info": {},
  "is_new": false,
  "trajectories_available": false,
  "references": [
    {
      "title": "Gemini Eval Documentation",
      "url": "https://storage.googleapis.com/deepmind-media/gemini/gemini_3_pro_model_evaluation.pdf",
      "type": "documentation"
    }
  ],
  "results": {
    "retail": {
      "pass_1": 85.3,
      "pass_2": null,
      "pass_3": null,
      "pass_4": null,
      "cost": null
    },
    "airline": {
      "pass_1": 73.0,
      "pass_2": null,
      "pass_3": null,
      "pass_4": null,
      "cost": null
    },
    "telecom": {
      "pass_1": 98.0,
      "pass_2": null,
      "pass_3": null,
      "pass_4": null,
      "cost": null
    }
  },
  "methodology": {
    "evaluation_date": "2025-11-18",
    "tau2_bench_version": "v0.1.3",
    "user_simulator": "Gemini-3.0-Pro",
    "notes": "τ2-bench results for Gemini use standard sierra framework with a prompt adjustment to provide instructions relevant to each environment. The user model uses Gemini with a system instruction. ",
    "verification": {
      "modified_prompts": true,
      "omitted_questions": false,
      "details": "Prompt adjustments made to provide instructions relevant to each environment. User model uses Gemini with system instruction."
    }
  },
  "model_release": {
    "release_date": "2025-11-18",
    "announcement_url": "https://blog.google/products/gemini/gemini-3/",
    "announcement_title": "Gemini 3: Introducing the latest Gemini AI model from Google"
  }
}