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

#ModelScore
1Kimi-K3Moonshot AI1494
2GLM-5.3Z.ai1493
3GLM-5.3-FlashZ.ai1490
4MiMo-V2.5-ProXiaomi MiMo1485
5DeepSeek-V4.1-FlashDeepSeek1483
6GLM-5.2Z.ai1480
7GLM-5.1Z.ai1473
8Kimi-K3.1-ProMoonshot AI1471
9Kimi-K2.6Moonshot AI1470
10DeepSeek-V4-ProDeepSeek1461
11Qwen3.8-27BQwen1456
12Kimi-K2.5Moonshot AI1453
13InklingThinking Machines1453
14NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16NVIDIA1452
15GLM-5Z.ai1452
16MiMo-V2.5Xiaomi MiMo1450
17Qwen3.5-397B-A17BQwen1448
18Hy3Tencent1447
19gemma-4-31B-itGoogle1447
20DeepSeek-V4-FlashDeepSeek1443
21GLM-4.7Z.ai1443
22Beam-1.0-501B-A23Breflection1442
23MiniMax-M3MiniMax1442
24GLM-4.6Z.ai1440
25gemma-4-26B-A4B-itGoogle1439
26Mistral-Medium-3.5-128BMistral AI1436
27Mistral-Large-4-1T-A52BMistral AI1434
28DeepSeek-V3.2DeepSeek1434
29Qwen3-235B-A22B-Instruct-2507Qwen1433
30LongCat-Flash-ChatMeituan LongCat1433
31Mistral-Large-3-675B-Instruct-2512Mistral AI1431
32GLM-4.5Z.ai1429
33DeepSeek-V3.2-ExpDeepSeek1429
34Qwen3-VL-235B-A22B-InstructQwen1428
35Qwen3-Next-80B-A3B-InstructQwen1428
36DeepSeek-V3.1-TerminusDeepSeek1426
37Inkling-SmallThinking Machines1425
38MiniMax-M2.7MiniMax1422
39Qwen3.5-122B-A10BQwen1421
40Inkling-2-SmallThinking Machines1421
41MiMo-V2-FlashXiaomi MiMo1420
42DeepSeek-V3.1DeepSeek1417
43Qwen3-235B-A22B-Thinking-2507Qwen1416
44DeepSeek-R1-0528DeepSeek1415
45Qwen3.5-27BQwen1414
46NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16NVIDIA1414
47Hy3-previewTencent1413
48Step-3.5-FlashStepFun1410
49Qwen3-VL-235B-A22B-ThinkingQwen1407
50Qwen3.5-35B-A3BQwen1400
51Qwen3-30B-A3B-Instruct-2507Qwen1398
52Muse-Glimmer-30Bmeta-models1395
53Qwen3-235B-A22BQwen1393
54NVIDIA-Nemotron-3-Super-120B-A12B-BF16NVIDIA1386
55GLM-4.5-AirZ.ai1379
56granite-4.2-30bIBM Granite1377
57MiniMax-M2.5MiniMax1372
58Qwen3-Coder-480B-A35B-InstructQwen1372
59Qwen3-Next-80B-A3B-ThinkingQwen1371
60GLM-4.6VZ.ai1368
61DeepSeek-V3-0324DeepSeek1365
62gpt-oss-120bOpenAI1363
63Ling-flash-2.0inclusionAI1362
64DeepSeek-R1DeepSeek1361
65MiniMax-M2MiniMax1357
66GLM-4.7-FlashZ.ai1356
67INTELLECT-3Prime Intellect1352
68Trinity-Large-PreviewArcee AI1350
69NVIDIA-Nemotron-3-Nano-30B-A3B-BF16NVIDIA1348
70Ring-flash-2.0inclusionAI1344
71Trinity-Large-ThinkingArcee AI1342
72gemma-3-27b-itGoogle1340
73NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16NVIDIA1340
74Mistral-Small-3.2-24B-Instruct-2506Mistral AI1335
75Qwen3-32BQwen1334
76GLM-4.5VZ.ai1334
77granite-4.2-8bIBM Granite1326
78QwQ-32BQwen1325
79Olmo-3.1-32B-InstructAi21322
80Qwen3-30B-A3BQwen1314
81DeepSeek-V3DeepSeek1312
82gemma-3-12b-itGoogle1309
83granite-4.2-3bIBM Granite1304
84Olmo-3-32B-ThinkAi21302
85granite-4.1-8bIBM Granite1294
86DeepSeek-V2.5-1210DeepSeek1289
87Molmo2-8BAi21287
88gemma-3n-E4B-itGoogle1284
89Llama-4-Maverick-17B-128E-InstructMeta Llama1281
90Mistral-Small-3.1-24B-Instruct-2503Mistral AI1278
91gpt-oss-20bOpenAI1274
92Olmo-3.1-32B-ThinkAi21272
93Qwen2.5-72B-InstructQwen1271
94DeepSeek-V2.5DeepSeek1270
95Llama-3.1-405B-InstructMeta Llama1269
96Llama-3.1-Nemotron-70B-InstructNVIDIA1266
97Llama-4-Scout-17B-16E-InstructMeta Llama1266
98Llama-3.3-70B-InstructMeta Llama1257
99gemma-3-4b-itGoogle1253
100Qwen2.5-Coder-32B-InstructQwen1251
101Llama-3.1-70B-InstructMeta Llama1241
102granite-4.0-h-smallIBM Granite1241
103Mistral-Small-24B-Instruct-2501Mistral AI1233
104Llama-3.1-Tulu-3-70BAi21220
105phi-4Microsoft1220
106DeepSeek-Coder-V2-InstructDeepSeek1207
107Nemotron-4-340B-InstructNVIDIA1201
108gemma-2-27b-itGoogle1198
109Meta-Llama-3-70B-InstructMeta Llama1195
110aya-expanse-32bCohere Labs1193
111Qwen2-72B-InstructQwen1191
112Llama-3.1-8B-InstructMeta Llama1175
113Llama-3.1-Tulu-3-8BAi21174
114gemma-2-9b-itGoogle1171
115Qwen1.5-110B-ChatQwen1168
116QwQ-32B-PreviewQwen1162
117aya-expanse-8bCohere Labs1156
118Mixtral-8x22B-Instruct-v0.1Mistral AI1150
119Qwen1.5-72B-ChatQwen1148
120granite-3.1-8b-instructIBM Granite1145
121granite-3.1-2b-instructIBM Granite1138
122Meta-Llama-3-8B-InstructMeta Llama1133
123Qwen1.5-32B-ChatQwen1130
124Phi-3-medium-4k-instructMicrosoft1126
125Mixtral-8x7B-Instruct-v0.1Mistral AI1115
126Qwen1.5-14B-ChatQwen1113
127gemma-2-2b-itGoogle1112
128Phi-3-small-8k-instructMicrosoft1100
129Llama-3.2-3B-InstructMeta Llama1095
130granite-3.0-8b-instructIBM Granite1092
131Phi-3-mini-4k-instructMicrosoft1087
132granite-3.0-2b-instructIBM Granite1073
133Llama-2-70b-chat-hfMeta Llama1073
134gemma-1.1-7b-itGoogle1071
135deepseek-llm-67b-chatDeepSeek1070
136Mistral-7B-Instruct-v0.2Mistral AI1067
137Qwen1.5-7B-ChatQwen1065
138SOLAR-10.7B-Instruct-v1.0Upstage1062
139Llama-2-13b-chat-hfMeta Llama1051
140Llama-3.2-1B-InstructMeta Llama1044
141gemma-7b-itGoogle1042
142Phi-3-mini-128k-instructMicrosoft1028
143Qwen-14B-ChatQwen1027
144Llama-2-7b-chat-hfMeta Llama1009
145gemma-1.1-2b-itGoogle1005
146OLMo-7B-InstructAi2993
147gemma-2b-itGoogle989
148Qwen1.5-4B-ChatQwen976
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.