LMArena Agent
Task-outcome score (0–100) from real agentic sessions judged by their users.
- Published by
- LMArena
- Category
- Agents & tools
- Index weight
- 0.5
- Models
- 40
- Data as of
- 9 Sept 2026
CC BY 4.0 — lmarena-ai/leaderboard-dataset on Hugging Face.
- 1Claude Fable 5.114.5
- 2GPT-6 Astra12.5
- 3Claude Opus 511.4
- 4Claude Opus 510.8
- 5Claude Fable 59.2
- 6GPT-5.6 Sol7.7
- 7Claude Opus 4.87.7
- 8Kimi K36.6
- 9Claude Sonnet 56.3
- 10GPT-5.55.3
- 11GLM-5.24.6
- 12DeepSeek V4 Pro4.4
- 13Gemini 3.8 Flash4
- 14Qwen3 83.9
- 15Grok 4.53.9
40 of 40
| # | |||||
|---|---|---|---|---|---|
| 1 |  Claude Fable 5.1Anthropic | 14.5 | – | 70.2 | – |
| 2 |  GPT-6 AstraOpenAI | 12.5 | – | 70.6 | – |
| 3 | Claude Opus 5Anthropic | 11.4 | – | 69.3 | – |
| 4 | Claude Opus 5Anthropic | 10.8 | – | 66.8 | – |
| 5 | Claude Fable 5Anthropic | 9.2 | – | 63.9 | – |
| 6 | GPT-5.6 SolOpenAI | 7.7 | – | 67.5 | – |
| 7 | Claude Opus 4.8Anthropic | 7.7 | – | 61.3 | – |
| 8 |  Kimi K3Moonshot AIopen | 6.6 | – | 62.7 | – |
| 9 | Claude Sonnet 5Anthropic | 6.3 | – | 60.5 | – |
| 10 | GPT-5.5OpenAI | 5.3 | – | 63.7 | – |
| 11 |  GLM-5.2Zhipu AIopen | 4.6 | – | 57.6 | – |
| 12 |  DeepSeek V4 ProDeepSeekopen | 4.4 | high, 0813 | 60.0 | – |
| 13 |  Gemini 3.8 FlashGoogle | 4 | – | 60.1 | – |
| 14 |  Qwen3 8Alibaba | 3.9 | – | 63.9 | – |
| 15 |  Grok 4.5xAI | 3.9 | Grok 4.5 | 63.4 | – |
| 16 | Grok 4.6xAI | 3.2 | – | 62.3 | – |
| 17 | GPT-5.5OpenAI | 3 | GPT 5.5 | 66.6 | – |
| 18 | GLM-5.3Zhipu AIopen | 2.6 | – | 59.9 | – |
| 19 | DeepSeek V4 FlashDeepSeekopen | 2.2 | high, 20260731 | 59.3 | – |
| 20 | GLM-5.3-FlashZhipu AIopen | 2 | GLM 5.3 Flash | 61.0 | – |
| 21 | GPT-5.4OpenAI | 1.6 | – | 58.8 | – |
| 22 | GPT-5.6 TerraOpenAI | 1.5 | – | 64.2 | – |
| 23 | GPT-5.6 LunaOpenAI | 1.1 | – | 59.3 | – |
| 24 | Qwen3.8-Flash-NextAlibabaopen | 0.8 | Qwen3.8 Flash Next | 59.3 | – |
| 25 | Qwen 3.8 27BAlibabaopen | 0.1 | Qwen 3.8 27B | 57.1 | – |
| 26 | DeepSeek V4 ProDeepSeekopen | -0.4 | DeepSeek V4 Pro | 59.7 | – |
| 27 | Claude Sonnet 4.6Anthropic | -1 | Claude Sonnet 4.6 | 60.3 | – |
| 28 |  Muse Spark 1.2Meta | -1.6 | – | 61.3 | – |
| 29 | Muse Spark 1.1Meta | -2.6 | Muse Spark 1.1 | 64.9 | – |
| 30 | Qwen3 7Alibaba | -3.1 | – | 58.9 | – |
| 31 | MiMo-V2.5-ProXiaomiopen | -5.1 | Mimo V2.5 Pro | 59.0 | – |
| 32 | Qwen3.7 PlusAlibaba | -5.2 | Qwen3.7 Plus | 59.7 | – |
| 33 | Gemini 3.1 ProGoogle | -5.3 | – | 63.9 | – |
| 34 | Gemini 3.6 FlashGoogle | -5.3 | – | 58.6 | – |
| 35 |  MiniMax-M3MiniMaxopen | -5.3 | Minimax M3 | 56.9 | – |
| 36 | Inkling-SmallThinking Machinesopen | -6.9 | – | 55.0 | – |
| 37 |  Mistral Medium 3.5Mistral AI | -9.3 | Mistral Medium 3.5 | 50.8 | – |
| 38 |  Solar Pro 4Upstage | -11.3 | Solar Pro 4 | – | – |
| 39 | MiniMax-M2.7MiniMaxopen | -13 | Minimax M2.7 | 56.5 | – |
| 40 | Gemini 3.5 Flash LiteGoogle | -13.4 | Gemini 3.5 Flash Lite | 53.5 | – |