{
  "generated": "2026-09-08",
  "benchmark": "tau-bench",
  "results": [
    {
      "system": "gpt-4o (function calling), tau-retail",
      "developer": "OpenAI",
      "value": 61.2,
      "date": "2024-06-17",
      "source": {
        "url": "https://arxiv.org/abs/2406.12045",
        "kind": "paper",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "split": "retail",
        "scaffold": "function calling (TC)",
        "notes": "Table 2 of the paper; user simulated by gpt-4."
      }
    },
    {
      "system": "gpt-4o (function calling), tau-airline",
      "developer": "OpenAI",
      "value": 35.2,
      "date": "2024-06-17",
      "source": {
        "url": "https://arxiv.org/abs/2406.12045",
        "kind": "paper",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "split": "airline",
        "scaffold": "function calling (TC)",
        "notes": "Table 2 of the paper; user simulated by gpt-4."
      }
    },
    {
      "system": "TC (claude-3-5-sonnet-20241022), retail",
      "developer": "Anthropic",
      "value": 69.2,
      "date": "2024-10-22",
      "source": {
        "url": "https://github.com/sierra-research/tau-bench",
        "kind": "official-leaderboard",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "split": "retail",
        "scaffold": "function calling (TC)",
        "notes": "Repository README leaderboard; date is the model's release date (README gives none)."
      }
    },
    {
      "system": "TC (claude-3-5-sonnet-20241022), airline",
      "developer": "Anthropic",
      "value": 46.0,
      "date": "2024-10-22",
      "source": {
        "url": "https://github.com/sierra-research/tau-bench",
        "kind": "official-leaderboard",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "split": "airline",
        "scaffold": "function calling (TC)",
        "notes": "Repository README leaderboard; date is the model's release date (README gives none)."
      }
    }
  ]
}