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

148 models ranked by LMArena score · updated Nov 11, 2026

#ModelScore
1GLM-5.3Z.ai1475
2Kimi-K3Moonshot AI1475
3GLM-5.3-FlashZ.ai1472
4GLM-5.2Z.ai1469
5MiMo-V2.5-ProXiaomi MiMo1464
6DeepSeek-V4.1-FlashDeepSeek1464
7GLM-5.1Z.ai1462
8Kimi-K3.1-ProMoonshot AI1459
9Kimi-K2.6Moonshot AI1455
10DeepSeek-V4-ProDeepSeek1451
11GLM-5Z.ai1446
12NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16NVIDIA1445
13Kimi-K2.5Moonshot AI1445
14InklingThinking Machines1443
15gemma-4-31B-itGoogle1443
16GLM-4.6Z.ai1440
17Hy3Tencent1439
18Qwen3.5-397B-A17BQwen1438
19Qwen3.8-27BQwen1437
20GLM-4.7Z.ai1435
21gemma-4-26B-A4B-itGoogle1434
22DeepSeek-V4-FlashDeepSeek1432
23GLM-4.5Z.ai1430
24Mistral-Large-3-675B-Instruct-2512Mistral AI1429
25MiniMax-M3MiniMax1428
26DeepSeek-R1-0528DeepSeek1428
27MiMo-V2.5Xiaomi MiMo1427
28DeepSeek-V3.2-ExpDeepSeek1425
29DeepSeek-V3.2DeepSeek1425
30Beam-1.0-501B-A23Breflection1424
31LongCat-Flash-ChatMeituan LongCat1423
32Mistral-Medium-3.5-128BMistral AI1421
33Qwen3-VL-235B-A22B-InstructQwen1420
34DeepSeek-V3.1DeepSeek1420
35DeepSeek-V3.1-TerminusDeepSeek1419
36Qwen3-235B-A22B-Instruct-2507Qwen1419
37Qwen3.5-122B-A10BQwen1417
38Qwen3-Next-80B-A3B-InstructQwen1417
39Qwen3-235B-A22B-Thinking-2507Qwen1415
40Inkling-SmallThinking Machines1414
41Mistral-Large-4-1T-A52BMistral AI1411
42MiMo-V2-FlashXiaomi MiMo1411
43Qwen3.5-27BQwen1409
44NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16NVIDIA1407
45MiniMax-M2.7MiniMax1405
46Inkling-2-SmallThinking Machines1403
47Step-3.5-FlashStepFun1402
48Hy3-previewTencent1401
49Qwen3-VL-235B-A22B-ThinkingQwen1401
50Qwen3.5-35B-A3BQwen1395
51Qwen3-235B-A22BQwen1394
52Muse-Glimmer-30Bmeta-models1389
53Qwen3-30B-A3B-Instruct-2507Qwen1384
54GLM-4.5-AirZ.ai1384
55NVIDIA-Nemotron-3-Super-120B-A12B-BF16NVIDIA1377
56GLM-4.6VZ.ai1377
57DeepSeek-V3-0324DeepSeek1375
58DeepSeek-R1DeepSeek1373
59Qwen3-Next-80B-A3B-ThinkingQwen1368
60gpt-oss-120bOpenAI1366
61granite-4.2-30bIBM Granite1362
62Ling-flash-2.0inclusionAI1361
63MiniMax-M2.5MiniMax1359
64gemma-3-27b-itGoogle1358
65Qwen3-Coder-480B-A35B-InstructQwen1357
66INTELLECT-3Prime Intellect1356
67GLM-4.7-FlashZ.ai1351
68NVIDIA-Nemotron-3-Nano-30B-A3B-BF16NVIDIA1348
69MiniMax-M2MiniMax1340
70Qwen3-32BQwen1340
71Trinity-Large-ThinkingArcee AI1340
72Trinity-Large-PreviewArcee AI1339
73Mistral-Small-3.2-24B-Instruct-2506Mistral AI1338
74gemma-3-12b-itGoogle1334
75Ring-flash-2.0inclusionAI1334
76GLM-4.5VZ.ai1333
77DeepSeek-V3DeepSeek1333
78NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16NVIDIA1329
79QwQ-32BQwen1329
80granite-4.2-8bIBM Granite1319
81Qwen3-30B-A3BQwen1317
82Olmo-3.1-32B-InstructAi21311
83gemma-3n-E4B-itGoogle1306
84Olmo-3-32B-ThinkAi21299
85DeepSeek-V2.5-1210DeepSeek1294
86granite-4.2-3bIBM Granite1292
87gemma-3-4b-itGoogle1291
88granite-4.1-8bIBM Granite1291
89gpt-oss-20bOpenAI1288
90Llama-4-Maverick-17B-128E-InstructMeta Llama1287
91Molmo2-8BAi21287
92Llama-3.1-405B-InstructMeta Llama1284
93Llama-3.1-Nemotron-70B-InstructNVIDIA1283
94Llama-4-Scout-17B-16E-InstructMeta Llama1279
95Mistral-Small-3.1-24B-Instruct-2503Mistral AI1277
96Llama-3.3-70B-InstructMeta Llama1274
97DeepSeek-V2.5DeepSeek1271
98Olmo-3.1-32B-ThinkAi21271
99Qwen2.5-72B-InstructQwen1269
100Llama-3.1-70B-InstructMeta Llama1261
101Llama-3.1-Tulu-3-70BAi21256
102granite-4.0-h-smallIBM Granite1241
103Mistral-Small-24B-Instruct-2501Mistral AI1234
104gemma-2-27b-itGoogle1231
105Qwen2.5-Coder-32B-InstructQwen1230
106Nemotron-4-340B-InstructNVIDIA1225
107aya-expanse-32bCohere Labs1224
108Meta-Llama-3-70B-InstructMeta Llama1221
109phi-4Microsoft1217
110gemma-2-9b-itGoogle1207
111Qwen2-72B-InstructQwen1203
112Llama-3.1-Tulu-3-8BAi21193
113DeepSeek-Coder-V2-InstructDeepSeek1191
114Llama-3.1-8B-InstructMeta Llama1187
115aya-expanse-8bCohere Labs1185
116Qwen1.5-110B-ChatQwen1175
117Qwen1.5-72B-ChatQwen1166
118Meta-Llama-3-8B-InstructMeta Llama1166
119Mixtral-8x22B-Instruct-v0.1Mistral AI1162
120QwQ-32B-PreviewQwen1162
121gemma-2-2b-itGoogle1156
122granite-3.1-8b-instructIBM Granite1150
123Phi-3-medium-4k-instructMicrosoft1138
124Qwen1.5-32B-ChatQwen1137
125Mixtral-8x7B-Instruct-v0.1Mistral AI1132
126Qwen1.5-14B-ChatQwen1128
127granite-3.1-2b-instructIBM Granite1127
128Llama-2-70b-chat-hfMeta Llama1115
129Phi-3-small-8k-instructMicrosoft1110
130Llama-3.2-3B-InstructMeta Llama1110
131deepseek-llm-67b-chatDeepSeek1105
132granite-3.0-8b-instructIBM Granite1096
133gemma-1.1-7b-itGoogle1094
134Mistral-7B-Instruct-v0.2Mistral AI1090
135Llama-2-13b-chat-hfMeta Llama1084
136Qwen1.5-7B-ChatQwen1084
137SOLAR-10.7B-Instruct-v1.0Upstage1083
138Phi-3-mini-4k-instructMicrosoft1080
139granite-3.0-2b-instructIBM Granite1080
140gemma-7b-itGoogle1056
141Llama-3.2-1B-InstructMeta Llama1055
142Llama-2-7b-chat-hfMeta Llama1053
143Qwen-14B-ChatQwen1051
144Phi-3-mini-128k-instructMicrosoft1050
145OLMo-7B-InstructAi21032
146gemma-1.1-2b-itGoogle1022
147gemma-2b-itGoogle1002
148Qwen1.5-4B-ChatQwen997
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.