Model leaderboard

Measured on BenchClub test sets

5 models · 4 test sets · 14 runs

<50%50–75%75–90%≥90%* Some answers cut off or failed

Arena scores (LMArena)

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

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