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

68 models in hard prompts ranked by LMArena score · updated Nov 11, 2026

#ModelScore
68Olmo-3.1-32B-ThinkAi21272
67gpt-oss-20bOpenAI1274
66Olmo-3-32B-ThinkAi21302
65granite-4.2-3bIBM Granite1304
64Qwen3-30B-A3BQwen1314
63QwQ-32BQwen1325
62granite-4.2-8bIBM Granite1326
61GLM-4.5VZ.ai1334
60Qwen3-32BQwen1334
59NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16NVIDIA1340
58Trinity-Large-ThinkingArcee AI1342
57NVIDIA-Nemotron-3-Nano-30B-A3B-BF16NVIDIA1348
56INTELLECT-3Prime Intellect1352
55GLM-4.7-FlashZ.ai1356
54MiniMax-M2MiniMax1357
53DeepSeek-R1DeepSeek1361
52Ling-flash-2.0inclusionAI1362
51gpt-oss-120bOpenAI1363
50GLM-4.6VZ.ai1368
49Qwen3-Next-80B-A3B-ThinkingQwen1371
Show all 68 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.