Agent Arena - Confirmed Success: leaderboard
Agent Arena submetric focused on real-world agent sessions that end in confirmed successful completion.
Metric: Confirmed Success (%). Source: arena.ai. Saturation forecast: Around December 2027. 51 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 (Max) | 17.29 | #3 (Claude Opus 5.5) |
| 2 | Claude Fable 5.1 (Max) | 16.19 | #9 (Claude Fable 5.1) |
| 3 | Claude Sonnet 5.5 (Max) | 15.83 | #11 (Claude Sonnet 5.5) |
| 4 | Gemini 4 Argon (High) | 15.64 | #8 (Gemini 4 Argon) |
| 5 | GPT-6.1 Sol (Max) | 14.74 | #7 (GPT-6.1 Sol) |
| 6 | GPT-6 Astra (Max) | 12.13 | #6 (GPT-6 Astra) |
| 7 | Claude Opus 5 (Max) | 10.42 | #12 (Claude Opus 5) |
| 8 | MiMo-V2.6-Pro | 8.03 | #42 |
| 9 | DeepSeek V4.1 Flash (Max) | 7.44 | #36 (DeepSeek V4.1 Flash) |
| 10 | GPT-6 Sol (Max) | 6.97 | #14 (GPT-6 Sol) |
| 11 | Kimi K3 (Max) | 6.46 | #30 (Kimi K3) |
| 12 | Hy4 preview | 5.96 | #38 |
| 13 | Claude Fable 5 (High) | 5.79 | #16 (Claude Fable 5) |
| 14 | MiMo-V2.6-Flash | 5.36 | #70 |
| 15 | GPT-5.6 Sol (xHigh) | 5.34 | #19 (GPT-5.6 Sol) |
No result here: #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=agent-arena-confirmed-success · How It Works · Data refreshed daily, snapshot 2026-10-20.