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

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

#ModelScore
139gemma-1.1-2b-itGoogle970
138Qwen1.5-4B-ChatQwen980
137Phi-3-mini-128k-instructMicrosoft984
136gemma-7b-itGoogle1001
135Llama-3.2-1B-InstructMeta Llama1007
134Llama-2-13b-chat-hfMeta Llama1030
133Mistral-7B-Instruct-v0.2Mistral AI1036
132Llama-2-7b-chat-hfMeta Llama1036
131Llama-2-70b-chat-hfMeta Llama1039
130gemma-1.1-7b-itGoogle1039
129Phi-3-mini-4k-instructMicrosoft1051
128Qwen1.5-7B-ChatQwen1055
127granite-3.0-2b-instructIBM Granite1064
126Phi-3-small-8k-instructMicrosoft1067
125granite-3.0-8b-instructIBM Granite1087
124Mixtral-8x7B-Instruct-v0.1Mistral AI1087
123Llama-3.2-3B-InstructMeta Llama1090
122Qwen1.5-14B-ChatQwen1094
121gemma-2-2b-itGoogle1096
120Phi-3-medium-4k-instructMicrosoft1108
Show all 139 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.