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

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

#ModelScore
86Llama-3.2-1B-InstructMeta Llama1030
85granite-3.0-2b-instructIBM Granite1053
84granite-3.0-8b-instructIBM Granite1063
83granite-3.1-2b-instructIBM Granite1099
82Llama-3.2-3B-InstructMeta Llama1105
81granite-3.1-8b-instructIBM Granite1108
80Mixtral-8x22B-Instruct-v0.1Mistral AI1130
79QwQ-32B-PreviewQwen1141
78Llama-3.1-8B-InstructMeta Llama1172
77Qwen2.5-Coder-32B-InstructQwen1222
76granite-4.0-h-smallIBM Granite1243
75Llama-3.1-70B-InstructMeta Llama1256
74gpt-oss-20bOpenAI1268
73granite-4.1-8bIBM Granite1268
72Qwen2.5-72B-InstructQwen1272
71Llama-3.3-70B-InstructMeta Llama1279
70Llama-4-Scout-17B-16E-InstructMeta Llama1280
69granite-4.2-3bIBM Granite1288
68Llama-4-Maverick-17B-128E-InstructMeta Llama1289
67Llama-3.1-405B-InstructMeta Llama1297
Show all 86 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.