Through an API

Open and closed models answer the same suite-standard 1.0 through their APIs, ranked together. 1 ranked model.

#ModelPass rate
1DeepSeek V4.1 Flash 1017Open weights100%

Model leaderboard

322 models in expert ranked by LMArena score · updated Nov 11, 2026

#ModelScore
322gemma-1.1-2b-itGoogle970
321Qwen1.5-4B-ChatQwen980
320Phi-3-mini-128k-instructMicrosoft984
319gemma-7b-itGoogle1001
318Llama-3.2-1B-InstructMeta Llama1007
317Llama-2-13b-chat-hfMeta Llama1030
316Mistral-7B-Instruct-v0.2Mistral AI1036
315Llama-2-7b-chat-hfMeta Llama1036
314Llama-2-70b-chat-hfMeta Llama1039
313gemma-1.1-7b-itGoogle1039
312Phi-3-mini-4k-instructMicrosoft1051
311Qwen1.5-7B-ChatQwen1055
310Gemini Pro (Dev API)Closed. Closed weights: ranked by its LMArena score, with no size or downloads to show.1059
309granite-3.0-2b-instructIBM Granite1064
308GPT-3.5 Turbo (0125)Closed. Closed weights: ranked by its LMArena score, with no size or downloads to show.1065
307Phi-3-small-8k-instructMicrosoft1067
306GPT-3.5 Turbo (1106)Closed. Closed weights: ranked by its LMArena score, with no size or downloads to show.1069
305granite-3.0-8b-instructIBM Granite1087
304Mixtral-8x7B-Instruct-v0.1Mistral AI1087
303Llama-3.2-3B-InstructMeta Llama1090
Show all 322 models

Score by size

Plot against
Best for its size

Scores: LMArena, CC BY 4.0. Measured on served versions, not local builds. A model with several served variants shows its best.