Leaderboards
Benchmarks
Every score carries its origin.
Arena Agent55Arena Agent · Recovery55Arena Agent · Steerability55Arena Agent · Task outcome55Arena Agent · Tool use55LiveBench Agentic Coding58SWE-bench Verified42
Arena Agent
Inverse-propensity score from live agent sessions — neither an Elo nor a percentageProvenance55 modelsscores Aug 11, 2026 – Sep 25, 2026
1Claude Fable 5.1AnthropicScore0.138ScoredSep 25, 20262GPT-6 AstraOpenAIScore0.109ScoredSep 25, 20263Claude Opus 5AnthropicScore0.098ScoredSep 25, 20264Claude Fable 5AnthropicScore0.083ScoredSep 25, 20265GPT-6 SolOpenAIScore0.077ScoredSep 25, 20266GPT-5.6 SolOpenAIScore0.062ScoredSep 25, 20267Claude Opus 4.7AnthropicScore0.057ScoredSep 5, 20268Claude Sonnet 5AnthropicScore0.048ScoredSep 25, 20269Claude Opus 4.6AnthropicScore0.047ScoredSep 5, 202610Kimi K3Moonshot AI2.8TScore0.046ScoredSep 25, 202611Hy4 previewTencent780BScore0.044ScoredSep 25, 202612Gemini 3.8 FlashGoogleScore0.041ScoredSep 25, 202613DeepSeek V4.1 FlashDeepSeek763BScore0.041ScoredSep 25, 202614GLM 5.2Z.ai753BScore0.04ScoredSep 25, 202615Muse Spark 1.3MetaScore0.039ScoredSep 25, 202616Grok 4.7xAIScore0.038ScoredSep 25, 202617GLM 5.3Z.aiScore0.029ScoredSep 25, 202618Qwen3.8 MaxQwenScore0.028ScoredSep 25, 202619Claude Opus 4.8AnthropicScore0.019ScoredSep 5, 202620DeepSeek V4 FlashDeepSeek291BScore0.018ScoredSep 15, 202621GPT-5.5OpenAIScore0.018ScoredSep 25, 202622Grok 4.5xAIScore0.016ScoredSep 25, 202623Grok 4.6xAIScore0.012ScoredSep 25, 202624GPT-5.6 TerraOpenAIScore0.008ScoredSep 25, 202625Kimi K2.7 CodeMoonshot AI1.1TScore0.008ScoredSep 5, 202626GPT-5.4OpenAIScore0.004ScoredSep 25, 202627GLM 5.3 FlashZ.ai321BScore0.002ScoredSep 25, 202628Kimi K2.6Moonshot AI1.1TScore−0.003ScoredSep 5, 202629Qwen3.8 FlashQwenScore−0.005ScoredSep 25, 202630GPT-5.6 LunaOpenAIScore−0.008ScoredSep 25, 202631DeepSeek V4 ProDeepSeek1.6TScore−0.011ScoredSep 25, 202632Gemini 3.7 FlashGoogleScore−0.012ScoredSep 25, 202633Qwen3.8 27BQwen27.8BScore−0.014ScoredSep 25, 202634Claude Sonnet 4.6AnthropicScore−0.015ScoredSep 15, 202635GLM 5.1Z.ai754BScore−0.022ScoredSep 5, 202636Gemini 3.6 FlashGoogleScore−0.027ScoredAug 11, 202637Muse Spark 1.2MetaScore−0.03ScoredSep 25, 202638Gemini 3.5 FlashGoogleScore−0.035ScoredSep 5, 202639Muse Spark 1.1MetaScore−0.039ScoredSep 25, 202640Qwen3.7 MaxQwenScore−0.044ScoredSep 25, 202641Hy3Tencent299BScore−0.056ScoredSep 25, 202642MiMo-V2.5-ProXiaomi1TScore−0.064ScoredSep 25, 202643MiniMax M3MiniMax427BScore−0.066ScoredSep 25, 202644Gemini 3.1 Pro Preview Custom ToolsGoogleScore−0.069ScoredSep 25, 202645Qwen3.7 PlusQwenScore−0.069ScoredSep 25, 202646Inkling SmallThinking Machines266BScore−0.096ScoredSep 25, 202647InklingThinking Machines952BScore−0.108ScoredSep 25, 202648Mistral Medium 3.5Mistral AI128BScore−0.116ScoredSep 25, 202649Grok Build 0.1xAIScore−0.129ScoredSep 5, 202650MiniMax M2.7MiniMax229BScore−0.14ScoredSep 15, 202651Nemotron 3 UltraNVIDIA561BScore−0.144ScoredSep 5, 202652Gemini 3.5 Flash LiteGoogleScore−0.153ScoredSep 15, 202653Solar Pro 4UpstageScore−0.162ScoredSep 25, 202654Grok 4.3xAIScore−0.203ScoredSep 5, 202655Gemma 4 31BGoogle31.3BScore−0.237ScoredSep 5, 2026
