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

338 models in multi-turn ranked by LMArena score · updated Nov 11, 2026

#ModelScore
338gemma-2b-itGoogle945
337gemma-1.1-2b-itGoogle958
336gemma-7b-itGoogle963
335Qwen1.5-4B-ChatQwen977
334Phi-3-mini-128k-instructMicrosoft989
333PaLM 2Closed. Closed weights: ranked by its LMArena score, with no size or downloads to show.994
332OLMo-7B-InstructAi21007
331Qwen-14B-ChatQwen1022
330Llama-2-7b-chat-hfMeta Llama1029
329Llama-3.2-1B-InstructMeta Llama1030
328gemma-1.1-7b-itGoogle1040
327SOLAR-10.7B-Instruct-v1.0Upstage1045
326Phi-3-mini-4k-instructMicrosoft1049
325Llama-2-13b-chat-hfMeta Llama1050
324granite-3.0-2b-instructIBM Granite1053
323Qwen1.5-7B-ChatQwen1061
322Mistral-7B-Instruct-v0.2Mistral AI1062
321granite-3.0-8b-instructIBM Granite1063
320Phi-3-small-8k-instructMicrosoft1068
319GPT-3.5 Turbo (1106)Closed. Closed weights: ranked by its LMArena score, with no size or downloads to show.1078
Show all 338 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.