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
1GLM-5.3Z.ai1493
2GLM-5.3-FlashZ.ai1490
3MiMo-V2.5-ProXiaomi MiMo1485
4DeepSeek-V4.1-FlashDeepSeek1483
5GLM-5.2Z.ai1480
6GLM-5.1Z.ai1473
7Kimi-K2.6Moonshot AI1470
8Qwen3.8-27BQwen1456
9Kimi-K2.5Moonshot AI1453
10InklingThinking Machines1453
11NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16NVIDIA1452
12GLM-5Z.ai1452
13MiMo-V2.5Xiaomi MiMo1450
14Qwen3.5-397B-A17BQwen1448
15Hy3Tencent1447
16gemma-4-31B-itGoogle1447
17GLM-4.7Z.ai1443
18Beam-1.0-501B-A23Breflection1442
19MiniMax-M3MiniMax1442
20GLM-4.6Z.ai1440
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.