LiveAgentBench: leaderboard
Metric: Share of the 374 LiveAgentBench tasks solved (%), pass@1, string-matched against closed answers, by agent products evaluated on their official websites with their own tools (and the AWorld framework with Claude 3.5 Sonnet); higher is better. Source: arxiv.org. Saturation forecast: Around November 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O4 Mini Deep Research | 27.54 | #219 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=liveagentbench · How It Works · Data refreshed daily, snapshot 2026-10-11.