Model leaderboard

Arena scores (LMArena)

189 models in math ranked by LMArena score · updated Nov 19, 2026

#ModelScore
189PaLM 2. Closed weights: ranked by its LMArena score, with no size or downloads to show.1049
188Gemini Pro. Closed weights: ranked by its LMArena score, with no size or downloads to show.1129
187Gemini Pro (Dev API). Closed weights: ranked by its LMArena score, with no size or downloads to show.1132
186GPT-3.5 Turbo (1106). Closed weights: ranked by its LMArena score, with no size or downloads to show.1141
185GPT-3.5 Turbo (0125). Closed weights: ranked by its LMArena score, with no size or downloads to show.1142
184Reka Flash 21B (20240226). Closed weights: ranked by its LMArena score, with no size or downloads to show.1156
183Reka Flash 21B (Online, 20240226). Closed weights: ranked by its LMArena score, with no size or downloads to show.1168
182Mistral Medium. Closed weights: ranked by its LMArena score, with no size or downloads to show.1180
181Claude 3 Haiku (20240307). Closed weights: ranked by its LMArena score, with no size or downloads to show.1188
180Reka Flash (20240904). Closed weights: ranked by its LMArena score, with no size or downloads to show.1195
179Mistral Large (2402). Closed weights: ranked by its LMArena score, with no size or downloads to show.1200
178Amazon Nova Micro 1.0. Closed weights: ranked by its LMArena score, with no size or downloads to show.1206
177Gemini 1.5 Flash-8B (001). Closed weights: ranked by its LMArena score, with no size or downloads to show.1207
176Claude 3 Sonnet (20240229). Closed weights: ranked by its LMArena score, with no size or downloads to show.1213
175GPT-4 (0613). Closed weights: ranked by its LMArena score, with no size or downloads to show.1217
174GLM-4 (0520). Closed weights: ranked by its LMArena score, with no size or downloads to show.1218
173Reka Core (20240904). Closed weights: ranked by its LMArena score, with no size or downloads to show.1222
172Amazon Nova Lite 1.0. Closed weights: ranked by its LMArena score, with no size or downloads to show.1227
171Gemini 1.5 Flash (001). Closed weights: ranked by its LMArena score, with no size or downloads to show.1229
170GPT-4 (0314). Closed weights: ranked by its LMArena score, with no size or downloads to show.1230
169Hunyuan Standard 256K. Closed weights: ranked by its LMArena score, with no size or downloads to show.1235
168Claude 3.5 Haiku (20241022). Closed weights: ranked by its LMArena score, with no size or downloads to show.1244
167Magistral Medium (2506). Closed weights: ranked by its LMArena score, with no size or downloads to show.1251
166Amazon Nova Pro 1.0. Closed weights: ranked by its LMArena score, with no size or downloads to show.1252
165Grok 2 mini (2024-08-13). Closed weights: ranked by its LMArena score, with no size or downloads to show.1265
164GPT-4o mini (2024-07-18). Closed weights: ranked by its LMArena score, with no size or downloads to show.1267
163GPT-4 Turbo Preview (0125). Closed weights: ranked by its LMArena score, with no size or downloads to show.1268
162Hunyuan Large Vision. Closed weights: ranked by its LMArena score, with no size or downloads to show.1268
161Gemini 1.5 Flash (002). Closed weights: ranked by its LMArena score, with no size or downloads to show.1269
160GPT-4 Turbo Preview (1106). Closed weights: ranked by its LMArena score, with no size or downloads to show.1269
159Gemini 1.5 Pro (001). Closed weights: ranked by its LMArena score, with no size or downloads to show.1270
158GPT-4 Turbo (2024-04-09). Closed weights: ranked by its LMArena score, with no size or downloads to show.1272
157Gemini Advanced (0514). Closed weights: ranked by its LMArena score, with no size or downloads to show.1272
156Claude 3 Opus (20240229). Closed weights: ranked by its LMArena score, with no size or downloads to show.1273
155Hunyuan Turbo (0110). Closed weights: ranked by its LMArena score, with no size or downloads to show.1273
154Hunyuan Standard (2025-02-10). Closed weights: ranked by its LMArena score, with no size or downloads to show.1274
153GPT-4.1 nano (2025-04-14). Closed weights: ranked by its LMArena score, with no size or downloads to show.1274
152GLM-4-Plus. Closed weights: ranked by its LMArena score, with no size or downloads to show.1275
151Hunyuan Large (2025-02-10). Closed weights: ranked by its LMArena score, with no size or downloads to show.1281
150Grok 2 (2024-08-13). Closed weights: ranked by its LMArena score, with no size or downloads to show.1283
149GPT-4o (2024-05-13). Closed weights: ranked by its LMArena score, with no size or downloads to show.1284
148GPT-4o (2024-08-06). Closed weights: ranked by its LMArena score, with no size or downloads to show.1285
147GLM-4-Plus (0111). Closed weights: ranked by its LMArena score, with no size or downloads to show.1287
146Hunyuan TurboS (20250226). Closed weights: ranked by its LMArena score, with no size or downloads to show.1294
145Qwen2.5-Plus (1127). Closed weights: ranked by its LMArena score, with no size or downloads to show.1298
144Yi Lightning. Closed weights: ranked by its LMArena score, with no size or downloads to show.1300
143Claude 3.5 Sonnet (20240620). Closed weights: ranked by its LMArena score, with no size or downloads to show.1303
142Step-2 16K Exp (202412). Closed weights: ranked by its LMArena score, with no size or downloads to show.1304
141Claude 3.5 Sonnet (20241022). Closed weights: ranked by its LMArena score, with no size or downloads to show.1307
140Gemini 2.0 Flash-Lite Preview (02-05). Closed weights: ranked by its LMArena score, with no size or downloads to show.1309
139Gemini 1.5 Pro (002). Closed weights: ranked by its LMArena score, with no size or downloads to show.1315
138Step-1o Turbo (202506). Closed weights: ranked by its LMArena score, with no size or downloads to show.1318
137GPT-5 nano (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1318
136Claude 3.7 Sonnet (20250219). Closed weights: ranked by its LMArena score, with no size or downloads to show.1319
135Qwen-Plus (0125). Closed weights: ranked by its LMArena score, with no size or downloads to show.1326
134Claude 3.7 Sonnet (Thinking 32K, 20250219). Closed weights: ranked by its LMArena score, with no size or downloads to show.1336
133GPT-4.1 mini (2025-04-14). Closed weights: ranked by its LMArena score, with no size or downloads to show.1343
132Gemini 2.0 Flash (001). Closed weights: ranked by its LMArena score, with no size or downloads to show.1352
131Mistral Medium 3 (2505). Closed weights: ranked by its LMArena score, with no size or downloads to show.1352
130o1-mini. Closed weights: ranked by its LMArena score, with no size or downloads to show.1358
129Amazon Nova 2 Lite. Closed weights: ranked by its LMArena score, with no size or downloads to show.1358
128Hunyuan TurboS (20250416). Closed weights: ranked by its LMArena score, with no size or downloads to show.1359
127Claude Sonnet 4 (20250514). Closed weights: ranked by its LMArena score, with no size or downloads to show.1359
126Gemini 2.5 Flash-Lite Preview (Thinking, 06-17). Closed weights: ranked by its LMArena score, with no size or downloads to show.1362
125Qwen2.5-Max. Closed weights: ranked by its LMArena score, with no size or downloads to show.1369
124GPT-4.1 (2025-04-14). Closed weights: ranked by its LMArena score, with no size or downloads to show.1370
123Grok 3 Mini Beta. Closed weights: ranked by its LMArena score, with no size or downloads to show.1372
122o1-preview. Closed weights: ranked by its LMArena score, with no size or downloads to show.1373
121Claude Opus 4 (20250514). Closed weights: ranked by its LMArena score, with no size or downloads to show.1373
120Gemini 2.5 Flash-Lite Preview (Non-Thinking, 09-2025). Closed weights: ranked by its LMArena score, with no size or downloads to show.1373
119o3-mini. Closed weights: ranked by its LMArena score, with no size or downloads to show.1374
118Claude Sonnet 4 (Thinking 32K, 20250514). Closed weights: ranked by its LMArena score, with no size or downloads to show.1375
117GPT-5 mini (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1378
116Grok 3 Mini (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1386
115Grok 4.3. Closed weights: ranked by its LMArena score, with no size or downloads to show.1387
114o1 (2024-12-17). Closed weights: ranked by its LMArena score, with no size or downloads to show.1388
113o4-mini (2025-04-16). Closed weights: ranked by its LMArena score, with no size or downloads to show.1389
112GPT-5.3 Chat. Closed weights: ranked by its LMArena score, with no size or downloads to show.1389
111Claude Opus 4 (Thinking 16K, 20250514). Closed weights: ranked by its LMArena score, with no size or downloads to show.1390
110Grok 3 Preview (02-24). Closed weights: ranked by its LMArena score, with no size or downloads to show.1391
109o3-mini (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1396
108Claude Haiku 4.5 (20251001). Closed weights: ranked by its LMArena score, with no size or downloads to show.1398
107GPT-5 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1398
106GPT-6 Sol (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1400
105amazon-nova-experimental-chat (26-01-10). Closed weights: ranked by its LMArena score, with no size or downloads to show.1401
104Grok 4 Fast (Reasoning). Closed weights: ranked by its LMArena score, with no size or downloads to show.1404
103ERNIE-5.0-Preview (1022). Closed weights: ranked by its LMArena score, with no size or downloads to show.1405
102GPT-5.4 nano (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1405
101ChatGPT-4o Latest (20250326). Closed weights: ranked by its LMArena score, with no size or downloads to show.1407
100Grok 4.1 Fast (Reasoning). Closed weights: ranked by its LMArena score, with no size or downloads to show.1407
99Qwen3.5-Flash. Closed weights: ranked by its LMArena score, with no size or downloads to show.1408
98GPT-5 Chat. Closed weights: ranked by its LMArena score, with no size or downloads to show.1408
97Mistral Medium 3.1 (2508). Closed weights: ranked by its LMArena score, with no size or downloads to show.1409
96GPT-6 Luna (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1410
95GPT-5.1. Closed weights: ranked by its LMArena score, with no size or downloads to show.1410
94Gemini 2.5 Flash. Closed weights: ranked by its LMArena score, with no size or downloads to show.1411
93ERNIE-5.0-Preview (1203). Closed weights: ranked by its LMArena score, with no size or downloads to show.1411
92GPT-4.5 Preview (2025-02-27). Closed weights: ranked by its LMArena score, with no size or downloads to show.1412
91Grok 4.7 (xHigh). Closed weights: ranked by its LMArena score, with no size or downloads to show.1412
90Hunyuan T1 (20250711). Closed weights: ranked by its LMArena score, with no size or downloads to show.1415
89Gemini 2.5 Flash Preview (09-2025). Closed weights: ranked by its LMArena score, with no size or downloads to show.1416
88Grok 4.1. Closed weights: ranked by its LMArena score, with no size or downloads to show.1417
87Grok 4.6 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1417
86Solar Pro 4. Closed weights: ranked by its LMArena score, with no size or downloads to show.1417
85GPT-5.4 mini (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1418
84GPT-5.2. Closed weights: ranked by its LMArena score, with no size or downloads to show.1418
83Grok 4 Fast (Non-Reasoning). Closed weights: ranked by its LMArena score, with no size or downloads to show.1419
82amazon-nova-experimental-chat (12-10). Closed weights: ranked by its LMArena score, with no size or downloads to show.1421
81Claude Sonnet 4.5 (20250929). Closed weights: ranked by its LMArena score, with no size or downloads to show.1421
80GPT-5.5 Instant. Closed weights: ranked by its LMArena score, with no size or downloads to show.1421
79Claude Opus 4.1 (20250805). Closed weights: ranked by its LMArena score, with no size or downloads to show.1422
78Grok 4 (0709). Closed weights: ranked by its LMArena score, with no size or downloads to show.1422
77Grok 4.1 (Thinking). Closed weights: ranked by its LMArena score, with no size or downloads to show.1423
76amazon-nova-experimental-chat (10-20). Closed weights: ranked by its LMArena score, with no size or downloads to show.1426
75o3 (2025-04-16). Closed weights: ranked by its LMArena score, with no size or downloads to show.1426
74Gemini 3.1 Flash-Lite Preview. Closed weights: ranked by its LMArena score, with no size or downloads to show.1428
73MiMo-V2-Omni. Closed weights: ranked by its LMArena score, with no size or downloads to show.1431
72Claude Opus 4.1 (Thinking 16K, 20250805). Closed weights: ranked by its LMArena score, with no size or downloads to show.1431
71amazon-nova-experimental-chat (11-10). Closed weights: ranked by its LMArena score, with no size or downloads to show.1432
70LongCat-Flash-Chat Exp (2602). Closed weights: ranked by its LMArena score, with no size or downloads to show.1432
69Claude Haiku 5.5 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1433
68Grok 4.20 Beta 1. Closed weights: ranked by its LMArena score, with no size or downloads to show.1433
67GPT-5.2 Chat (20260210). Closed weights: ranked by its LMArena score, with no size or downloads to show.1434
66Gemini 3.5 Flash-Lite. Closed weights: ranked by its LMArena score, with no size or downloads to show.1434
65ERNIE 5.0 (0110). Closed weights: ranked by its LMArena score, with no size or downloads to show.1436
64Qwen3-Max (2025-09-23). Closed weights: ranked by its LMArena score, with no size or downloads to show.1436
63Dola Seed 2.0 Pro. Closed weights: ranked by its LMArena score, with no size or downloads to show.1439
62amazon-nova-experimental-chat (26-02-10). Closed weights: ranked by its LMArena score, with no size or downloads to show.1439
61GPT-5.2 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1440
60Grok 4.20 Multi-Agent Beta (0309). Closed weights: ranked by its LMArena score, with no size or downloads to show.1441
59Qwen3-Max-Preview. Closed weights: ranked by its LMArena score, with no size or downloads to show.1447
58GPT-5.1 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1447
57MiMo-V2-Pro. Closed weights: ranked by its LMArena score, with no size or downloads to show.1447
56Gemini 3 Flash (Minimal). Closed weights: ranked by its LMArena score, with no size or downloads to show.1447
55Qwen3.6-Plus. Closed weights: ranked by its LMArena score, with no size or downloads to show.1447
54Claude Sonnet 4.5 (High, 32K, 20250929). Closed weights: ranked by its LMArena score, with no size or downloads to show.1449
53GPT-6.1 Terra (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1450
52Gemini 2.5 Pro. Closed weights: ranked by its LMArena score, with no size or downloads to show.1450
51Grok 4.20 Beta (Reasoning, 0309). Closed weights: ranked by its LMArena score, with no size or downloads to show.1454
50GPT-6 Astra (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1456
49GPT-5.4. Closed weights: ranked by its LMArena score, with no size or downloads to show.1457
48GPT-5.6 Luna (xHigh). Closed weights: ranked by its LMArena score, with no size or downloads to show.1457
47Muse Spark. Closed weights: ranked by its LMArena score, with no size or downloads to show.1457
46Grok 4.5. Closed weights: ranked by its LMArena score, with no size or downloads to show.1457
45Claude Opus 4.5 (20251101). Closed weights: ranked by its LMArena score, with no size or downloads to show.1458
44GPT-5.6 Terra (xHigh). Closed weights: ranked by its LMArena score, with no size or downloads to show.1461
43Claude Sonnet 4.6. Closed weights: ranked by its LMArena score, with no size or downloads to show.1461
42Qwen3.6-Max-Preview. Closed weights: ranked by its LMArena score, with no size or downloads to show.1462
41Claude Opus 4.5 (High, 32K, 20251101). Closed weights: ranked by its LMArena score, with no size or downloads to show.1463
40Step 5 Preview. Closed weights: ranked by its LMArena score, with no size or downloads to show.1466
39Qwen3.7-Plus. Closed weights: ranked by its LMArena score, with no size or downloads to show.1466
38Claude Sonnet 5 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1466
37Claude Opus 4.8. Closed weights: ranked by its LMArena score, with no size or downloads to show.1470
36GPT-5.6 Sol (xHigh). Closed weights: ranked by its LMArena score, with no size or downloads to show.1471
35Gemini 3 Flash. Closed weights: ranked by its LMArena score, with no size or downloads to show.1473
34Muse Spark 1.2 (xHigh). Closed weights: ranked by its LMArena score, with no size or downloads to show.1473
33Qwen3.5-Max-Preview. Closed weights: ranked by its LMArena score, with no size or downloads to show.1475
32Gemini 3 Pro. Closed weights: ranked by its LMArena score, with no size or downloads to show.1475
31GPT-6.1 Sol (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1475
30GPT-5.5 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1476
29Muse Spark 1.1. Closed weights: ranked by its LMArena score, with no size or downloads to show.1482
28ERNIE 5.1. Closed weights: ranked by its LMArena score, with no size or downloads to show.1482
27Qwen3.7-Max-Preview. Closed weights: ranked by its LMArena score, with no size or downloads to show.1485
26Gemini 3.1 Pro Preview. Closed weights: ranked by its LMArena score, with no size or downloads to show.1485
25Claude Opus 4.8 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1485
24GPT-5.5. Closed weights: ranked by its LMArena score, with no size or downloads to show.1486
23Gemini 3.5 Flash (Medium). Closed weights: ranked by its LMArena score, with no size or downloads to show.1487
22GPT-5.4 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1488
21Claude Opus 4.7. Closed weights: ranked by its LMArena score, with no size or downloads to show.1489
20Claude Fable 5.2 (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1491
19Muse Wonder (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1493
18Claude Sonnet 5.5 (xHigh). Closed weights: ranked by its LMArena score, with no size or downloads to show.1494
17Muse Spark 1.3 (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1494
16Claude Opus 4.7 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1498
15Qwen3.8-Max. Closed weights: ranked by its LMArena score, with no size or downloads to show.1505
14Gemini 3.6 Flash (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1505
13Gemini 3.5 Flash (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1505
12Gemini 3.7 Flash (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1506
11Claude Opus 4.6. Closed weights: ranked by its LMArena score, with no size or downloads to show.1510
10Gemini 4 Hydrogen (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1515
9Claude Fable 5 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1516
8Claude Opus 4.6 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1517
7Claude Opus 5.5 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1523
6Gemini 3.8 Flash (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1525
5Claude Opus 5 (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1532
4Claude Opus 5 (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1532
3Claude Fable 5.1 (Max). Closed weights: ranked by its LMArena score, with no size or downloads to show.1533
2Gemini 4 Argon (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1543
1Gemini 4 Neon (High). Closed weights: ranked by its LMArena score, with no size or downloads to show.1548
Show the top 20

Scores: LMArena, CC BY 4.0. Measured on served versions, not local builds. A model with several served variants shows its best.