LMArena Agent Arena Confirmed Success
Agents · 2026-06-04
Signed percentage-point Agent Arena diagnostic signal for sessions where the user explicitly confirmed task completion. Values may be negative when a model is below the leaderboard baseline; higher values indicate more reliably successful agent sessions.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 22.1 |
| Claude Sonnet 5.5 | 21.7 |
| Claude Fable 5.1 | 13.9 |
| GPT-6.1 Sol | 13.7 |
| Gemini 4 Argon | 13.5 |
| GPT-6 Astra | 13.2 |
| Muse Wonder Full | 11.8 |
| Claude Opus 5 | 11.6 |
| Muse Wonder | 7.7 |
| DeepSeek V4.1 Flash 1017 | 7.1 |
| GPT-6 Sol | 6.9 |
| MiMo-V2.6-Pro | 6.9 |
| MiMo-V2.6-Flash | 6.2 |
| Grok 4.7 | 6.0 |
| DeepSeek V4.1 Flash | 5.8 |
| Kimi K3 | 5.2 |
| Hy4 Preview | 5.1 |
| Muse Spark 1.3 | 4.3 |
| Qwen3.8-Max | 3.8 |
| Claude Fable 5 | 3.4 |
| Gemini 3.8 Flash | 2.9 |
| GLM-5.2 | 2.8 |
| GLM-5.3 | 2.6 |
| GLM-5.3 Flash | 2.4 |
| Kimi K3.1 Pro | 2.0 |
| GPT-5.6 Sol | 1.9 |
| Claude Haiku 5.5 | 1.7 |
| Qwen3.8-Flash-Next | 1.6 |
| GPT-6 Luna | 0.5 |
| Mistral Large 4 Preview | -0.3 |
| Step 5 Preview | -0.5 |
| Sonnet 5 | -1.1 |
| Grok 4.6 | -1.8 |
| Qwen3.8 27B | -2.6 |
| Gemini 3.7 Flash | -2.9 |
| DeepSeek V4 Pro 0813 | -3.2 |
| GPT-5.6 Terra | -4.4 |
| Muse Spark 1.1 | -6.2 |
| Muse Spark 1.2 | -9.2 |
| Qwen3.7-Max (source unspecified) | -9.9 |