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

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

#ModelScore
31gemma-3-4b-itGoogle1255
30gemma-3n-E4B-itGoogle1276
29Llama-4-Scout-17B-16E-InstructMeta Llama1280
28Llama-4-Maverick-17B-128E-InstructMeta Llama1289
27GLM-4.5VZ.ai1333
26gemma-3-12b-itGoogle1334
25gemma-3-27b-itGoogle1345
24GLM-4.6VZ.ai1358
23Qwen3-VL-235B-A22B-ThinkingQwen1388
22Qwen3.5-35B-A3BQwen1390
21Muse-Glimmer-30Bmeta-models1397
20Inkling-2-SmallThinking Machines1408
19Qwen3.5-27BQwen1410
18NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16NVIDIA1415
17Qwen3.5-122B-A10BQwen1416
16Inkling-SmallThinking Machines1417
15Mistral-Large-4-1T-A52BMistral AI1421
14Qwen3-VL-235B-A22B-InstructQwen1428
13Qwen3.8-27BQwen1436
12MiniMax-M3MiniMax1439
Show all 31 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.