Instruction Following
LMArena Text Leaderboard
Arena (arena.ai, formerly LMArena / Chatbot Arena) ranks models by millions of blind, head-to-head human votes on real prompts. The Arena Score shown here is a Bradley-Terry rating computed with Arena's default style control, so higher means people preferred that model's answers more often — it's a relative rating, not a percentage, so compare it only to other models on this board.
Source: lmarena188 open models ranked+220 proprietaryData through Oct 2026
All models ranked on LMArena Text
Proprietary / closed models are shown dimmed — you can't run them locally, but they show where the open field stands.
| # | Model | Score | VRAM @ Q4 |
|---|---|---|---|
| 1 | Gemini 4 Carbon (High) · proprietary | 1527.5 | API only |
| 2 | Gemini 4 Argon (High) · proprietary | 1520.3 | API only |
| 3 | Claude Opus 5.5 (High) · proprietary | 1508.7 | API only |
| 4 | Muse Wonder (Max) · proprietary | 1507.1 | API only |
| 5 | Claude Fable 5.1 (Max) · proprietary | 1504.4 | API only |
| 6 | Claude Fable 5 (High) · proprietary | 1503.9 | API only |
| 7 | Claude Opus 4.6 (High) · proprietary | 1503.2 | API only |
| 8 | Claude Opus 4.7 (High) · proprietary | 1501.0 | API only |
| 9 | Claude Opus 4.6 · proprietary | 1496.1 | API only |
| 10 | Gemini 3.8 Flash (High) · proprietary | 1495.9 | API only |
| 11 | Muse Spark 1.3 (Max) · proprietary | 1495.0 | API only |
| 12 | Claude Opus 4.7 · proprietary | 1493.7 | API only |
| 13 | Muse Spark 1.2 (xHigh) · proprietary | 1492.2 | API only |
| 14 | Claude Opus 5 (High) · proprietary | 1491.6 | API only |
| 15 | Claude Opus 5 (Max) · proprietary | 1490.5 | API only |
| 16 | Muse Spark 1.1 · proprietary | 1490.2 | API only |
| 17 | Gemini 3.7 Flash (High) · proprietary | 1489.6 | API only |
| 18 | Muse Spark · proprietary | 1489.0 | API only |
| 19 | Kimi K3 · 2779.9B | 1488.2 | ~1674 GB |
| 20 | Gemini 3.1 Pro Preview · proprietary | 1486.9 | API only |
| 21 | Gemini 3 Pro · proprietary | 1485.4 | API only |
| 22 | GPT 5.6 Sol (xHigh) · proprietary | 1484.9 | API only |
| 23 | Gemini 3.6 Flash (High) · proprietary | 1483.2 | API only |
| 24 | Qwen3.8 Max · proprietary | 1482.4 | API only |
| 25 | Claude Opus 4.8 (High) · proprietary | 1481.7 | API only |
| 26 | GPT 5.5 (High) · proprietary | 1481.7 | API only |
| 27 | GPT 6.1 Sol (Max) · proprietary | 1481.3 | API only |
| 28 | GLM 5.3 · 753.3B | 1480.6 | ~456 GB |
| 29 | MiMo V2.6 Pro RL · 1020B | 1479.7 | ~613 GB |
| 30 | Gemini 3.5 Flash (High) · proprietary | 1477.6 | API only |
| 31 | Claude Sonnet 5.5 (xHigh) · proprietary | 1477.1 | API only |
| 32 | GPT 5.5 · proprietary | 1476.7 | API only |
| 33 | GLM 5.3 Flash · 321.3B | 1476.5 | ~195 GB |
| 34 | Gemini 3.5 Flash (Medium) · proprietary | 1476.4 | API only |
| 35 | DeepSeek V4.1 Flash · 763.2B | 1476.2 | ~458 GB |
| 36 | GPT 5.2 Chat (20260210) · proprietary | 1475.7 | API only |
| 37 | GPT 6 Astra (Max) · proprietary | 1475.6 | API only |
| 38 | GLM 5.2 · 753.3B | 1475.3 | ~456 GB |
| 39 | GPT 5.4 (High) · proprietary | 1475.3 | API only |
| 40 | Grok 4.20 Beta 1 · proprietary | 1474.6 | API only |
| 41 | Claude Opus 4.8 · proprietary | 1474.4 | API only |
| 42 | Qwen3.7 Max Preview · proprietary | 1473.7 | API only |
| 43 | Claude Opus 4.5 (High, 32K, 20251101) · proprietary | 1473.7 | API only |
| 44 | GPT 5.5 Instant · proprietary | 1473.2 | API only |
| 45 | Gemini 3 Flash · proprietary | 1472.7 | API only |
| 46 | Claude Sonnet 4.6 · proprietary | 1472.0 | API only |
| 47 | Grok 4.20 Beta (Reasoning, 0309) · proprietary | 1471.7 | API only |
| 48 | Grok 4.20 Multi Agent Beta (0309) · proprietary | 1470.7 | API only |
| 49 | Claude Opus 4.5 (20251101) · proprietary | 1469.7 | API only |
| 50 | GPT 5.6 Terra (xHigh) · proprietary | 1467.3 | API only |
| 51 | ERNIE 5.1 · proprietary | 1467.1 | API only |
| 52 | MiMo V2.5 Pro · 1023.2B | 1466.6 | ~614 GB |
| 53 | Grok 4.5 · proprietary | 1465.4 | API only |
| 54 | GLM 5.1 · 753.9B | 1465.4 | ~457 GB |
| 55 | DeepSeek V4 Pro 0813 · 1650.5B | 1465.1 | ~991 GB |
| 56 | Grok 4.1 (Thinking) · proprietary | 1465.0 | API only |
| 57 | GPT 5.4 · proprietary | 1465.0 | API only |
| 58 | Qwen3.5 Max Preview · proprietary | 1464.6 | API only |
| 59 | Claude Sonnet 5 (High) · proprietary | 1461.7 | API only |
| 60 | Kimi K2.6 · 1026.9B | 1461.0 | ~620 GB |
| 61 | Qwen3.6 Max Preview · proprietary | 1460.2 | API only |
| 62 | Grok 4.1 · proprietary | 1459.0 | API only |
| 63 | Gemini 3 Flash (Minimal) · proprietary | 1458.4 | API only |
| 64 | GPT 6 Sol (Max) · proprietary | 1458.3 | API only |
| 65 | DeepSeek V4 Pro · 1598.8B | 1457.8 | ~960 GB |
| 66 | GLM 5 · 753.9B | 1457.2 | ~457 GB |
| 67 | Claude Sonnet 4.5 (High, 32K, 20250929) · proprietary | 1456.3 | API only |
| 68 | Dola Seed 2.0 Pro · proprietary | 1456.1 | API only |
| 69 | GPT 5.1 (High) · proprietary | 1455.7 | API only |
| 70 | Qwen3.7 Plus · proprietary | 1455.7 | API only |
| 71 | Gemini 3.5 Flash Lite · proprietary | 1455.7 | API only |
| 72 | Grok 4.6 (High) · proprietary | 1455.5 | API only |
| 73 | Claude Sonnet 4.5 (20250929) · proprietary | 1455.1 | API only |
| 74 | Hy3 · 298.8B | 1454.5 | ~180 GB |
| 75 | GPT 5.6 Luna (xHigh) · proprietary | 1453.0 | API only |
| 76 | Gemma 4 31B IT · 31.3B | 1452.3 | ~20 GB |
| 77 | GPT 5.3 Chat · proprietary | 1450.1 | API only |
| 78 | Claude Opus 4.1 (Thinking 16K, 20250805) · proprietary | 1449.8 | API only |
| 79 | Kimi K2.5 · 1026.9B | 1449.7 | ~620 GB |
| 80 | ERNIE 5.0 Preview (1203) · proprietary | 1448.5 | API only |
| 81 | Step 5 Preview · proprietary | 1448.4 | API only |
| 82 | Claude Opus 4.1 (20250805) · proprietary | 1448.0 | API only |
| 83 | MiMo V2 Pro · proprietary | 1447.8 | API only |
| 84 | MiMo V2.6 Flash RL · 309B | 1447.6 | ~186 GB |
| 85 | GPT 5.4 mini (High) · proprietary | 1447.2 | API only |
| 86 | ERNIE 5.0 (0110) · proprietary | 1446.1 | API only |
| 87 | Gemini 2.5 Pro · proprietary | 1445.4 | API only |
| 88 | GPT 4.5 Preview (2025) (Feb 27) · proprietary | 1444.6 | API only |
| 89 | Qwen3.6 Plus · proprietary | 1443.0 | API only |
| 90 | ChatGPT 4o Latest (20250326) · proprietary | 1442.7 | API only |
| 91 | Inkling · 952.4B | 1442.2 | ~572 GB |
| 92 | Grok 4.3 · proprietary | 1441.6 | API only |
| 93 | Qwen3.5 397B A17B · 403.4B | 1441.4 | ~242 GB |
| 94 | GLM 4.7 · 358.3B | 1441.3 | ~216 GB |
| 95 | Grok 4.7 (xHigh) · proprietary | 1441.2 | API only |
| 96 | GPT 6 Luna (Max) · proprietary | 1440.9 | API only |
| 97 | MiniMax M3 · 427.0B | 1439.2 | ~257 GB |
| 98 | DeepSeek V4 Flash · 290.9B | 1438.7 | ~175 GB |
| 99 | GPT 5.1 · proprietary | 1438.5 | API only |
| 100 | Gemma 4 26B A4B IT · 25.8B | 1437.6 | ~16 GB |
| 101 | GPT 5.2 (High) · proprietary | 1437.3 | API only |
| 102 | LongCat Flash Chat Exp (2602) · proprietary | 1436.4 | API only |
| 103 | Beam 1.0 501B A23B · 501.3B | 1435.7 | ~301 GB |
| 104 | Qwen3.8 27B · 27.8B | 1435.6 | ~17 GB |
| 105 | GPT 5.2 · proprietary | 1435.6 | API only |
| 106 | GPT 5 (High) · proprietary | 1434.7 | API only |
| 107 | Qwen3 Max Preview · proprietary | 1434.3 | API only |
| 108 | GLM 5V Turbo · proprietary | 1434.0 | API only |
| 109 | MiMo V2.5 · 310.8B | 1433.9 | ~187 GB |
| 110 | Gemini 3.1 Flash Lite Preview · proprietary | 1432.9 | API only |
| 111 | Mistral Large 4 · proprietary | 1432.0 | API only |
| 112 | O3 (2025) (Apr 16) · proprietary | 1431.7 | API only |
| 113 | MiMo V2 Omni · proprietary | 1430.7 | API only |
| 114 | Grok 4.1 Fast (Reasoning) · proprietary | 1430.3 | API only |
| 115 | Kimi K2 Thinking · 1026.4B | 1430.0 | ~620 GB |
| 116 | Amazon Nova Experimental Chat (26) (Feb 10) · proprietary | 1426.7 | API only |
| 117 | NVIDIA Nemotron 3 Ultra 550B A55B BF16 · 560.5B | 1426.5 | ~370 GB |
| 118 | GPT 5 Chat · proprietary | 1426.4 | API only |
| 119 | Mistral Medium 3.5 128B · 127.7B | 1426.4 | ~78 GB |
| 120 | Claude Opus 4 (Thinking 16K, 20250514) · proprietary | 1426.2 | API only |
| 121 | DeepSeek V3.2 Exp · 685.4B | 1425.1 | ~415 GB |
| 122 | Muse Glimmer 30B · 29.8B | 1424.9 | ~18 GB |
| 123 | Claude Haiku 5.5 (High) · proprietary | 1424.8 | API only |
| 124 | DeepSeek V3.2 · 685.4B | 1424.6 | ~415 GB |
| 125 | GLM 4.6 · 356.8B | 1424.3 | ~215 GB |
| 126 | Qwen3 Max (2025) (Sep 23) · proprietary | 1423.5 | API only |
| 127 | Qwen3 235B A22B Instruct 2507 · 235.1B | 1422.2 | ~142 GB |
| 128 | DeepSeek R1 0528 · 684.5B | 1421.5 | ~415 GB |
| 129 | Kimi K2 Instruct 0905 · 1026.4B | 1418.5 | ~620 GB |
| 130 | Grok 4 Fast (Non Reasoning) · proprietary | 1418.4 | API only |
| 131 | ERNIE 5.0 Preview (1022) · proprietary | 1418.3 | API only |
| 132 | Kimi K2 Instruct · 1026.4B | 1417.8 | ~620 GB |
| 133 | DeepSeek V3.1 · 684.5B | 1416.9 | ~415 GB |
| 134 | Qwen3.5 122B A10B · 125.1B | 1416.5 | ~75 GB |
| 135 | DeepSeek V3.1 Terminus · 684.5B | 1416.1 | ~415 GB |
| 136 | MiniMax M2.7 · 228.7B | 1415.0 | ~138 GB |
| 137 | GPT 4.1 (2025) (Apr 14) · proprietary | 1414.6 | API only |
| 138 | Claude Haiku 4.5 (20251001) · proprietary | 1414.5 | API only |
| 139 | Mistral Large 3 675B Instruct 2512 · 675B | 1414.3 | ~446 GB |
| 140 | Claude Opus 4 (20250514) · proprietary | 1414.3 | API only |
| 141 | Qwen3 VL 235B A22B Instruct · 235.7B | 1413.3 | ~142 GB |
| 142 | Amazon Nova Experimental Chat (26) (Jan 10) · proprietary | 1412.8 | API only |
| 143 | NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 · 124.5B | 1412.3 | ~82 GB |
| 144 | Grok 3 Preview (02.24) · proprietary | 1411.2 | API only |
| 145 | Grok 4 (0709) · proprietary | 1411.0 | API only |
| 146 | GLM 4.5 · 358.3B | 1410.9 | ~216 GB |
| 147 | Gemini 2.5 Flash · proprietary | 1409.2 | API only |
| 148 | Qwen3.5 27B · 27.8B | 1408.8 | ~17 GB |
| 149 | Hy3 Preview · 298.8B | 1408.8 | ~180 GB |
| 150 | Mistral Medium 3.1 (2508) · proprietary | 1407.8 | API only |
| 151 | Inkling Small · 266.0B | 1405.5 | ~160 GB |
| 152 | Grok 4 Fast (Reasoning) · proprietary | 1405.3 | API only |
| 153 | Gemini 2.5 Flash Preview (09 2025) · proprietary | 1403.0 | API only |
| 154 | O1 (2024) (Dec 17) · proprietary | 1402.3 | API only |
| 155 | Qwen3 235B A22B · 235.1B | 1402.1 | ~142 GB |
| 156 | Claude Sonnet 4 (Thinking 32K, 20250514) · proprietary | 1401.9 | API only |
| 157 | GPT 5.4 nano (High) · proprietary | 1401.2 | API only |
| 158 | LongCat Flash Chat · 561.9B | 1400.8 | ~1236 GB |
| 159 | Qwen3 235B A22B Thinking 2507 · 235.1B | 1400.1 | ~142 GB |
| 160 | Qwen3 Next 80B A3B Instruct · 81.3B | 1398.5 | ~49 GB |
| 161 | DeepSeek R1 · 684.5B | 1398.1 | ~415 GB |
| 162 | Qwen3.5 Flash · proprietary | 1396.0 | API only |
| 163 | DeepSeek v3 0324 · 684.5B | 1395.2 | ~415 GB |
| 164 | Qwen3 VL 235B A22B Thinking · 235.7B | 1394.7 | ~142 GB |
| 165 | Hunyuan Vision 1.5 Thinking · proprietary | 1394.1 | API only |
| 166 | Qwen3.5 35B A3B · 36.0B | 1393.8 | ~22 GB |
| 167 | Step 3.5 Flash · 199.4B | 1393.0 | ~132 GB |
| 168 | Amazon Nova Experimental Chat (12.10) · proprietary | 1392.9 | API only |
| 169 | MiMo v2 Flash · 309.8B | 1391.5 | ~186 GB |
| 170 | O4 Mini (2025) (Apr 16) · proprietary | 1391.0 | API only |
| 171 | Claude Sonnet 4 (20250514) · proprietary | 1390.9 | API only |
| 172 | MiniMax M2.5 · 228.7B | 1390.6 | ~138 GB |
| 173 | GPT 5 mini (High) · proprietary | 1389.8 | API only |
| 174 | O1 Preview · proprietary | 1388.6 | API only |
| 175 | Claude 3.7 Sonnet (Thinking 32K, 20250219) · proprietary | 1388.3 | API only |
| 176 | Solar Pro 4 · proprietary | 1387.8 | API only |
| 177 | Qwen3 Coder 480B A35B Instruct · 480.2B | 1387.6 | ~289 GB |
| 178 | Hunyuan T1 (20250711) · proprietary | 1387.3 | API only |
| 179 | Mistral Medium 3 (2505) · proprietary | 1387.3 | API only |
| 180 | MiniMax M2.1 · 228.7B | 1384.5 | ~138 GB |
| 181 | Hunyuan TurboS (20250416) · proprietary | 1383.1 | API only |
| 182 | GPT 4.1 mini (2025) (Apr 14) · proprietary | 1382.8 | API only |
| 183 | Qwen3 30B A3B Instruct 2507 · 30.5B | 1382.3 | ~19 GB |
| 184 | Gemini 2.5 Flash Lite Preview (Non Thinking, 09 2025) · proprietary | 1379.4 | API only |
| 185 | Trinity Large Preview · 398.6B | 1378.5 | ~798 GB |
| 186 | GLM 4.6V · 107.7B | 1378.4 | ~65 GB |
| 187 | Gemini 2.5 Flash Lite Preview (Thinking, 06.17) · proprietary | 1374.7 | API only |
| 188 | Claude 3.5 Sonnet (20241022) · proprietary | 1374.3 | API only |
| 189 | Qwen2.5 Max · proprietary | 1373.8 | API only |
| 190 | GLM 4.5 Air · 110.5B | 1373.0 | ~67 GB |
| 191 | Claude 3.7 Sonnet (20250219) · proprietary | 1372.5 | API only |
| 192 | Qwen3 Next 80B A3B Thinking · 81.3B | 1369.2 | ~49 GB |
| 193 | Trinity Large Thinking · 398.6B | 1367.5 | ~798 GB |
| 194 | Gemma 3 27B IT · 27.4B | 1365.3 | ~18 GB |
| 195 | GLM 4.7 Flash · 31.2B | 1364.7 | ~20 GB |
| 196 | MiniMax M1 80k · 456.1B | 1363.8 | ~913 GB |
| 197 | Amazon Nova Experimental Chat (11.10) · proprietary | 1363.8 | API only |
| 198 | Grok 3 Mini (High) · proprietary | 1363.7 | API only |
| 199 | O3 Mini (High) · proprietary | 1363.6 | API only |
| 200 | Gemini 2.0 Flash (001) · proprietary | 1360.0 | API only |
| 201 | NVIDIA Nemotron 3 Super 120B A12B BF16 · 123.6B | 1359.9 | ~75 GB |
| 202 | DeepSeek v3 · 684.5B | 1358.4 | ~415 GB |
| 203 | Grok 3 Mini Beta · proprietary | 1357.9 | API only |
| 204 | Mistral Small 3.2 24B Instruct 2506 · 24.0B | 1356.2 | ~15 GB |
| 205 | INTELLECT 3 · 106.9B | 1355.5 | ~65 GB |
| 206 | C4ai Command A 03 2025 · 111.1B | 1353.7 | ~73 GB |
| 207 | Gemini 2.0 Flash Lite Preview (02.05) · proprietary | 1353.5 | API only |
| 208 | NVIDIA Nemotron 3.5 Lightning 30B A3B BF16 · 31.6B | 1353.3 | ~19 GB |
| 209 | GLM 4.5V · 107.7B | 1352.1 | ~65 GB |
| 210 | GPT OSS 120B · 116.8B | 1351.7 | ~71 GB |
| 211 | Gemini 1.5 Pro (002) · proprietary | 1351.3 | API only |
| 212 | Step3 · 321.0B | 1349.0 | ~212 GB |
| 213 | Hunyuan TurboS (20250226) · proprietary | 1348.8 | API only |
| 214 | Amazon Nova Experimental Chat (10.20) · proprietary | 1348.2 | API only |
| 215 | O3 Mini · proprietary | 1348.1 | API only |
| 216 | Llama 3 1 Nemotron Ultra 253B V1 · 253.4B | 1347.6 | ~557 GB |
| 217 | Qwen3 32B · 32.8B | 1346.8 | ~20 GB |
| 218 | Amazon Nova Experimental Chat (10.09) · proprietary | 1346.2 | API only |
| 219 | GPT 4o (2024) (May 13) · proprietary | 1346.1 | API only |
| 220 | Qwen Plus (0125) · proprietary | 1345.9 | API only |
| 221 | MiniMax M2 · 228.7B | 1343.5 | ~138 GB |
| 222 | Claude 3.5 Sonnet (20240620) · proprietary | 1343.3 | API only |
| 223 | Ling Flash 2.0 · 102.9B | 1343.3 | ~62 GB |
| 224 | Llama 3 3 Nemotron Super 49B V1 5 · 49.9B | 1343.1 | ~33 GB |
| 225 | Mercury 2 · proprietary | 1342.8 | API only |
| 226 | GLM 4 Plus (0111) · proprietary | 1342.7 | API only |
| 227 | Gemma 3 12B IT · 12.2B | 1341.7 | ~8.0 GB |
| 228 | Hunyuan Turbo (0110) · proprietary | 1340.7 | API only |
| 229 | Granite 4.2 30B · 29.3B | 1340.1 | ~18 GB |
| 230 | GPT 5 nano (High) · proprietary | 1337.7 | API only |
| 231 | O1 Mini · proprietary | 1336.8 | API only |
| 232 | Gemini Advanced (0514) · proprietary | 1336.0 | API only |
| 233 | Grok 2 (2024) (Aug 13) · proprietary | 1335.7 | API only |
| 234 | GPT 4o (2024) (Aug 06) · proprietary | 1335.6 | API only |
| 235 | QwQ 32B · 32.8B | 1335.6 | ~21 GB |
| 236 | Llama 3.1 405B Instruct · 405.9B | 1335.5 | ~268 GB |
| 237 | Amazon Nova 2 Lite · proprietary | 1335.0 | API only |
| 238 | Step 2 16K Exp (202412) · proprietary | 1333.5 | API only |
| 239 | Olmo 3.1 32B Instruct · 32.2B | 1329.3 | ~20 GB |
| 240 | Yi Lightning · proprietary | 1328.3 | API only |
| 241 | Llama 3 3 Nemotron Super 49B V1 · 49.9B | 1327.7 | ~33 GB |
| 242 | Llama 4 Maverick 17B 128E Instruct · 401.6B | 1327.0 | ~265 GB |
| 243 | Qwen3 30B A3B · 30.5B | 1326.7 | ~19 GB |
| 244 | Hunyuan Large (2025) (Feb 10) · proprietary | 1326.3 | API only |
| 245 | Claude 3.5 Haiku (20241022) · proprietary | 1324.8 | API only |
| 246 | GPT 4 Turbo (2024) (Apr 09) · proprietary | 1324.5 | API only |
| 247 | Gemini 1.5 Pro (001) · proprietary | 1324.1 | API only |
| 248 | DeepSeek V2.5 1210 · 235.7B | 1323.4 | ~144 GB |
| 249 | Ring Flash 2.0 · 102.9B | 1322.4 | ~62 GB |
| 250 | GPT 4.1 nano (2025) (Apr 14) · proprietary | 1322.3 | API only |
| 251 | Claude 3 Opus (20240229) · proprietary | 1322.2 | API only |
| 252 | Llama 4 Scout 17B 16E Instruct · 108.6B | 1321.4 | ~72 GB |
| 253 | Molmo2 8B · 8.7B | 1321.4 | ~5.8 GB |
| 254 | Step 1o Turbo (202506) · proprietary | 1319.8 | API only |
| 255 | GLM 4 Plus · proprietary | 1319.4 | API only |
| 256 | Qwen Max (0919) · proprietary | 1318.1 | API only |
| 257 | Llama 3.3 70B Instruct · 70.6B | 1318.0 | ~47 GB |
| 258 | GPT 4o mini (2024) (Jul 18) · proprietary | 1317.6 | API only |
| 259 | Gemma 3n E4B IT · 7.8B | 1317.5 | ~5.2 GB |
| 260 | GPT OSS 20B · 20.9B | 1317.3 | ~13 GB |
| 261 | Qwen2.5 Plus (1127) · proprietary | 1314.6 | API only |
| 262 | Mistral Large Instruct 2407 · 122.6B | 1314.4 | ~81 GB |
| 263 | Athene v2 Chat · 72.7B | 1314.1 | ~45 GB |
| 264 | NVIDIA Nemotron 3 Nano 30B A3B BF16 · 31.6B | 1313.6 | ~19 GB |
| 265 | GPT 4 Turbo Preview (0125) · proprietary | 1313.3 | API only |
| 266 | GPT 4 Turbo Preview (1106) · proprietary | 1312.9 | API only |
| 267 | Hunyuan Standard (2025) (Feb 10) · proprietary | 1311.3 | API only |
| 268 | Gemini 1.5 Flash (002) · proprietary | 1309.3 | API only |
| 269 | Grok 2 mini (2024) (Aug 13) · proprietary | 1308.4 | API only |
| 270 | DeepSeek V2.5 · 235.7B | 1307.2 | ~144 GB |
| 271 | Athene 70B (0725) · proprietary | 1306.6 | API only |
| 272 | Olmo 3 32B Think · 32.2B | 1305.9 | ~20 GB |
| 273 | Mercury · proprietary | 1305.7 | API only |
| 274 | Mistral Large Instruct 2411 · 122.6B | 1305.6 | ~75 GB |
| 275 | Magistral Medium (2506) · proprietary | 1304.8 | API only |
| 276 | Granite 4.1 8B · 8.8B | 1304.0 | ~5.9 GB |
| 277 | Gemma 3 4B IT · 4.3B | 1303.2 | ~2.8 GB |
| 278 | Qwen2.5 72B Instruct · 72.7B | 1302.9 | ~45 GB |
| 279 | Mistral Small 3.1 24B Instruct 2503 · 24.0B | 1302.6 | ~15 GB |
| 280 | Llama 3.1 Nemotron 70B Instruct HF · 70.6B | 1298.7 | ~43 GB |
| 281 | Hunyuan Large Vision · proprietary | 1293.7 | API only |
| 282 | Llama 3.1 70B Instruct · 70.6B | 1293.4 | ~47 GB |
| 283 | Amazon Nova Pro 1.0 · proprietary | 1290.1 | API only |
| 284 | Granite 4.2 3B · 3.7B | 1290.0 | ~2.7 GB |
| 285 | Granite 4.2 8B · 8.8B | 1289.7 | ~5.9 GB |
| 286 | Gemma 2 27B IT · 27.2B | 1289.5 | ~18 GB |
| 287 | AI21 Jamba Large 1.5 · 398.6B | 1289.5 | ~877 GB |
| 288 | Reka Core (20240904) · proprietary | 1288.0 | API only |
| 289 | GPT 4 (0314) · proprietary | 1287.9 | API only |
| 290 | Granite 4.0 H Small · 32.2B | 1287.3 | ~20 GB |
| 291 | Llama 3 1 Nemotron 51B Instruct · 51B | 1286.7 | ~112 GB |
| 292 | Gemini 1.5 Flash (001) · proprietary | 1286.5 | API only |
| 293 | Olmo 3.1 32B Think · 32.2B | 1286.4 | ~65 GB |
| 294 | Llama 3.1 Tulu 3 70B · 70.6B | 1286.1 | ~43 GB |
| 295 | Claude 3 Sonnet (20240229) · proprietary | 1281.4 | API only |
| 296 | Gemma 2 9B IT SimPO · 9.2B | 1280.4 | ~6.5 GB |
| 297 | Nemotron 4 340B Instruct · proprietary | 1277.2 | API only |
| 298 | Meta Llama 3 70B Instruct · 70.6B | 1276.6 | ~47 GB |
| 299 | C4ai Command R Plus 08 2024 · 103.8B | 1276.3 | ~69 GB |
| 300 | GPT 4 (0613) · proprietary | 1276.3 | API only |
| 301 | Mistral Small 24B Instruct 2501 · 23.6B | 1274.4 | ~15 GB |
| 302 | GLM 4 (0520) · proprietary | 1273.4 | API only |
| 303 | Reka Flash (20240904) · proprietary | 1272.4 | API only |
| 304 | Qwen2.5 Coder 32B Instruct · 32.8B | 1270.6 | ~21 GB |
| 305 | Aya Expanse 32B · 32.3B | 1267.3 | ~21 GB |
| 306 | Gemma 2 9B IT · 9.2B | 1267.0 | ~6.1 GB |
| 307 | DeepSeek Coder v2 Instruct · 235.7B | 1265.4 | ~144 GB |
| 308 | Qwen2 72B Instruct · 72.7B | 1261.9 | ~45 GB |
| 309 | C4ai Command R Plus · 103.8B | 1261.8 | ~69 GB |
| 310 | Claude 3 Haiku (20240307) · proprietary | 1261.7 | API only |
| 311 | Amazon Nova Lite 1.0 · proprietary | 1260.4 | API only |
| 312 | Gemini 1.5 Flash 8B (001) · proprietary | 1258.8 | API only |
| 313 | Phi 4 · 14.7B | 1256.3 | ~9.5 GB |
| 314 | OLMo 2 0325 32B Instruct · 32.2B | 1251.7 | ~20 GB |
| 315 | C4ai Command R 08 2024 · 32.3B | 1250.5 | ~21 GB |
| 316 | Mistral Large (2402) · proprietary | 1242.6 | API only |
| 317 | Amazon Nova Micro 1.0 · proprietary | 1240.7 | API only |
| 318 | AI21 Jamba Mini 1.5 · 51.6B | 1239.8 | ~113 GB |
| 319 | Ministral 8B Instruct 2410 · 8.0B | 1237.6 | ~5.4 GB |
| 320 | Gemini Pro (Dev API) · proprietary | 1236.9 | API only |
| 321 | Qwen1.5 110B Chat · 111.2B | 1234.2 | ~68 GB |
| 322 | Reka Flash 21B (Online, 20240226) · proprietary | 1233.6 | API only |
| 323 | Qwen1.5 72B Chat · 72.3B | 1233.6 | ~49 GB |
| 324 | Hunyuan Standard 256K · proprietary | 1233.2 | API only |
| 325 | Mixtral 8x22B Instruct v0.1 · 140.6B | 1229.7 | ~85 GB |
| 326 | C4ai Command R V01 · 35.0B | 1226.9 | ~23 GB |
| 327 | Reka Flash 21B (20240226) · proprietary | 1226.9 | API only |
| 328 | GPT 3.5 Turbo (0125) · proprietary | 1225.6 | API only |
| 329 | Meta Llama 3 8B Instruct · 8.0B | 1223.7 | ~5.3 GB |
| 330 | Gemini Pro · proprietary | 1223.7 | API only |
| 331 | Aya Expanse 8B · 8.0B | 1223.1 | ~18 GB |
| 332 | Mistral Medium · proprietary | 1222.8 | API only |
| 333 | Llama 3.1 Tulu 3 8B · 8.0B | 1220.5 | ~5.4 GB |
| 334 | Zephyr Orpo 141B A35b v0.1 · 140.6B | 1213.1 | ~85 GB |
| 335 | Yi 1.5 34B Chat · 34.4B | 1212.9 | ~21 GB |
| 336 | Llama 3.1 8B Instruct · 8.0B | 1211.2 | ~5.3 GB |
| 337 | Granite 3.1 8B Instruct · 8.2B | 1208.6 | ~5.5 GB |
| 338 | GPT 3.5 Turbo (1106) · proprietary | 1204.3 | API only |
| 339 | Qwen1.5 32B Chat · 32.5B | 1204.1 | ~20 GB |
| 340 | Gemma 2 2B IT · 2.6B | 1200.1 | ~1.7 GB |
| 341 | Phi 3 Medium 4k Instruct · 14.0B | 1198.0 | ~9.1 GB |
| 342 | Mixtral 8x7B Instruct v0.1 · 46.7B | 1197.2 | ~29 GB |
| 343 | DBRX Instruct Preview · proprietary | 1195.6 | API only |
| 344 | Qwen1.5 14B Chat · 14.2B | 1191.1 | ~10 GB |
| 345 | Internlm2 5 20B Chat · 19.9B | 1190.8 | ~13 GB |
| 346 | Deepseek Llm 67B Chat · 67B | 1185.0 | ~41 GB |
| 347 | WizardLM 70B · proprietary | 1184.8 | API only |
| 348 | Yi 34B Chat · 34.4B | 1183.8 | ~21 GB |
| 349 | Granite 3.0 8B Instruct · 8.2B | 1183.2 | ~5.5 GB |
| 350 | OpenChat 3.5 · proprietary | 1183.1 | API only |
| 351 | Openchat 3.5 0106 · 7.2B | 1182.8 | ~4.9 GB |
| 352 | Gemma 1.1 7B IT · 8.5B | 1182.6 | ~19 GB |
| 353 | Snowflake Arctic Instruct · 478.6B | 1180.1 | ~958 GB |
| 354 | Granite 3.1 2B Instruct · 2.5B | 1178.9 | ~2.0 GB |
| 355 | Tulu 2 DPO 70B · proprietary | 1177.7 | API only |
| 356 | OpenHermes 2.5 Mistral 7B · 7B | 1176.0 | ~4.8 GB |
| 357 | Vicuna 33B V1.3 · 33B | 1173.0 | ~22 GB |
| 358 | Phi 3 Small 8k Instruct · 7.4B | 1171.1 | ~15 GB |
| 359 | Starling LM 7B Beta · 7.2B | 1171.0 | ~4.9 GB |
| 360 | Llama 2 70B Chat HF · 69.0B | 1170.7 | ~152 GB |
| 361 | Starling LM 7B Alpha · 7.2B | 1167.5 | ~4.9 GB |
| 362 | Llama 3.2 3B Instruct · 3.2B | 1166.8 | ~2.1 GB |
| 363 | Nous Hermes 2 Mixtral 8x7B DPO · 46.7B | 1164.5 | ~29 GB |
| 364 | Granite 3.0 2B Instruct · 2.6B | 1156.7 | ~2.0 GB |
| 365 | Llama 2 70B SteerLM Chat · proprietary | 1154.8 | API only |
| 366 | QwQ 32B Preview · 32.8B | 1154.2 | ~21 GB |
| 367 | SOLAR 10.7B Instruct v1.0 · 10.7B | 1152.5 | ~7.1 GB |
| 368 | Dolphin 2.2.1 Mistral 7B · proprietary | 1152.3 | API only |
| 369 | MPT 30B Chat · proprietary | 1150.8 | API only |
| 370 | WizardLM 13B · proprietary | 1149.6 | API only |
| 371 | Mistral 7B Instruct v0.2 · 7.2B | 1149.5 | ~4.9 GB |
| 372 | Falcon 180B Chat · proprietary | 1148.2 | API only |
| 373 | Qwen1.5 7B Chat · 7.7B | 1143.9 | ~6.0 GB |
| 374 | Phi 3 Mini 4k Instruct · 3.8B | 1143.2 | ~3.4 GB |
| 375 | Vicuna 13B · proprietary | 1141.6 | API only |
| 376 | Llama 2 13B Chat HF · 13.0B | 1141.5 | ~8.6 GB |
| 377 | Qwen 14B Chat · 14.2B | 1139.4 | ~9.3 GB |
| 378 | PaLM 2 · proprietary | 1139.2 | API only |
| 379 | Gemma 7B IT · 8.5B | 1137.8 | ~19 GB |
| 380 | CodeLlama 34B Instruct HF · 33.7B | 1136.9 | ~21 GB |
| 381 | Zephyr 7B Beta · 7.2B | 1130.9 | ~4.9 GB |
| 382 | Phi 3 Mini 128k Instruct · 3.8B | 1129.9 | ~3.4 GB |
| 383 | Guanaco 33B · proprietary | 1127.6 | API only |
| 384 | Zephyr 7B Alpha · proprietary | 1127.0 | API only |
| 385 | StripedHyena Nous 7B · 7.6B | 1121.7 | ~17 GB |
| 386 | CodeLlama 70B Instruct HF · 69.0B | 1119.2 | ~42 GB |
| 387 | Gemma 1.1 2B IT · 2.5B | 1116.7 | ~1.6 GB |
| 388 | Vicuna 7B · proprietary | 1115.4 | API only |
| 389 | SmolLM2 1.7B Instruct · 1.7B | 1114.8 | ~1.7 GB |
| 390 | Llama 3.2 1B Instruct · 1.2B | 1111.1 | ~0.8 GB |
| 391 | Mistral 7B Instruct v0.1 · 7.2B | 1110.4 | ~4.9 GB |
| 392 | Llama 2 7B Chat HF · 6.7B | 1108.0 | ~4.5 GB |
| 393 | Gemma 2B IT · 2.5B | 1093.7 | ~1.6 GB |
| 394 | Qwen1.5 4B Chat · proprietary | 1091.2 | API only |
| 395 | OLMo 7B Instruct · proprietary | 1073.7 | API only |
| 396 | Koala 13B · proprietary | 1071.0 | API only |
| 397 | Alpaca 13B · proprietary | 1070.3 | API only |
| 398 | GPT4All 13B Snoozy · proprietary | 1067.8 | API only |
| 399 | MPT 7B Chat · proprietary | 1063.5 | API only |
| 400 | Chatglm3 6B · 6.2B | 1056.6 | ~4.1 GB |
| 401 | RWKV 4 Raven 14B · proprietary | 1042.4 | API only |
| 402 | Chatglm2 6B · 6B | 1024.8 | ~4.0 GB |
| 403 | OpenAssistant Pythia 12B · proprietary | 1023.6 | API only |
| 404 | ChatGLM 6B · proprietary | 995.9 | API only |
| 405 | FastChat T5 3B · proprietary | 992.6 | API only |
| 406 | Dolly 2.0 12B · proprietary | 982.6 | API only |
| 407 | LLaMA 13B · proprietary | 975.2 | API only |
| 408 | Stablelm Tuned Alpha 7B · 7B | 953.3 | ~4.6 GB |
Score vs model size
Which models give the most quality for their size — the ones worth running locally.
- Llama 3.2 1B Instruct, 1B, score 1111.1 — on the efficiency frontier (best score at its size or smaller).
- SmolLM2 1.7B Instruct, 2B, score 1114.8 — on the efficiency frontier (best score at its size or smaller).
- Gemma 1.1 2B IT, 3B, score 1116.7 — on the efficiency frontier (best score at its size or smaller).
- Granite 3.1 2B Instruct, 3B, score 1178.9 — on the efficiency frontier (best score at its size or smaller).
- Gemma 2 2B IT, 3B, score 1200.1 — on the efficiency frontier (best score at its size or smaller).
- Granite 4.2 3B, 4B, score 1290.0 — on the efficiency frontier (best score at its size or smaller).
- Gemma 3 4B IT, 4B, score 1303.2 — on the efficiency frontier (best score at its size or smaller).
- Gemma 3n E4B IT, 8B, score 1317.5 — on the efficiency frontier (best score at its size or smaller).
- Molmo2 8B, 9B, score 1321.4 — on the efficiency frontier (best score at its size or smaller).
- Gemma 3 12B IT, 12B, score 1341.7 — on the efficiency frontier (best score at its size or smaller).
- Mistral Small 3.2 24B Instruct 2506, 24B, score 1356.2 — on the efficiency frontier (best score at its size or smaller).
- Gemma 4 26B A4B IT, 26B, score 1437.6 — on the efficiency frontier (best score at its size or smaller).
- Gemma 4 31B IT, 31B, score 1452.3 — on the efficiency frontier (best score at its size or smaller).
- Hy3, 299B, score 1454.5 — on the efficiency frontier (best score at its size or smaller).
- GLM 5.3 Flash, 321B, score 1476.5 — on the efficiency frontier (best score at its size or smaller).
- GLM 5.3, 753B, score 1480.6 — on the efficiency frontier (best score at its size or smaller).
- Kimi K3, 2.8T, score 1488.2 — on the efficiency frontier (best score at its size or smaller).
LMArena Text: frequently asked questions
- What is the best open LLM on LMArena Text?
- Kimi K3 is the top open model on LMArena Text, scoring 1488.2. Among all models tested — including proprietary ones — it ranks #19. The top model overall is Gemini 4 Carbon (High) (Google) at 1527.5.
- What's the best LMArena Text model you can run on a 24 GB GPU?
- Gemma 4 31B IT is the highest-scoring open model that fits in 24 GB at 4-bit quantization (~20 GB), scoring 1452.3 on LMArena Text.
- What's the best LMArena Text model you can run on a 12 GB GPU?
- Gemma 3 12B IT is the highest-scoring open model that fits in 12 GB at 4-bit quantization (~8.0 GB), scoring 1341.7 on LMArena Text.
- Can open models match proprietary models on LMArena Text?
- Not quite on LMArena Text: the strongest proprietary model (Gemini 4 Carbon (High)) scores 1527.5, ahead of the best open model (Kimi K3) at 1488.2 — but you can run the open one yourself.
Scores aggregated from lmarena. llmrun does not run this benchmark — see the source for methodology, or the about benchmarks for what it measures.