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

28 models in math ranked by LMArena score · updated Nov 11, 2026

#ModelScore
1GLM-5.3-FlashZ.ai1503
2Kimi-K3Moonshot AI1496
3InklingThinking Machines1481
4DeepSeek-V4.1-FlashDeepSeek1479
5Kimi-K2.6Moonshot AI1475
6gemma-4-26B-A4B-itGoogle1470
7Kimi-K2.5Moonshot AI1469
8gemma-4-31B-itGoogle1462
9Inkling-SmallThinking Machines1459
10Mistral-Large-4-1T-A52BMistral AI1456
11Qwen3.8-27BQwen1455
12Qwen3.5-397B-A17BQwen1453
13MiMo-V2.5Xiaomi MiMo1436
14Qwen3.5-122B-A10BQwen1431
15Qwen3.5-27BQwen1429
16Qwen3-VL-235B-A22B-InstructQwen1427
17MiniMax-M3MiniMax1423
18NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16NVIDIA1413
19Muse-Glimmer-30Bmeta-models1412
20Qwen3-VL-235B-A22B-ThinkingQwen1410
21Qwen3.5-35B-A3BQwen1403
22GLM-4.5VZ.ai1354
23gemma-3-27b-itGoogle1312
24gemma-3-12b-itGoogle1307
25Llama-4-Maverick-17B-128E-InstructMeta Llama1300
26Llama-4-Scout-17B-16E-InstructMeta Llama1287
27gemma-3n-E4B-itGoogle1250
28gemma-3-4b-itGoogle1239
Show the top 20

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.