{
  "generated": "2026-09-08",
  "benchmark": "agentbench",
  "results": [
    {
      "system": "gpt-4 (0613)",
      "developer": "OpenAI",
      "value": 4.01,
      "date": "2023-08-07",
      "source": {
        "url": "https://arxiv.org/abs/2308.03688",
        "kind": "paper",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "notes": "Overall AgentBench score (Table 3); weighted average, not a percentage."
      }
    },
    {
      "system": "gpt-3.5-turbo (0613)",
      "developer": "OpenAI",
      "value": 2.32,
      "date": "2023-08-07",
      "source": {
        "url": "https://arxiv.org/abs/2308.03688",
        "kind": "paper",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "notes": "Overall AgentBench score (Table 3)."
      }
    },
    {
      "system": "claude-3 opus",
      "developer": "Anthropic",
      "value": 3.11,
      "date": "2025-10-04",
      "source": {
        "url": "https://arxiv.org/abs/2308.03688",
        "kind": "paper",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "notes": "Overall score in Table 3 of arXiv v3 (October 2025 revision, which added later models)."
      }
    }
  ]
}