LMArena Text Arena
LMArena · text-2026-08-27-51d09b2e834b · Blind pairwise human preferences from LMArena's text arena.
Model ranking
Publisher benchmark →Source | |||||||
|---|---|---|---|---|---|---|---|
| #1 | Claude Opus 5Anthropic | 1504.75 | — | 15,398 | claude-opus-5-max; 95% CI [1498.76198432, 1510.73188918]; votes 15398; rank 1unknown | Original sourcewinner eligible | Third-party benchmark |
| #2 | Claude Fable 5Anthropic | 1494.57 | — | 25,824 | claude-fable-5; 95% CI [1489.51776584, 1499.62569530]; votes 25824; rank 5unknown | Original sourcewinner eligible | Third-party benchmark |
| #3 | Kimi-K3Moonshot AI | 1476.06 | — | 16,586 | kimi-k3-max; 95% CI [1470.44912152, 1481.67231358]; votes 16586; rank 16unknown | Original sourcewinner eligible | Third-party benchmark |
| #4 | GLM-5.2Z.ai | 1466.95 | — | 32,495 | glm-5.2-max; 95% CI [1462.19875405, 1471.69805536]; votes 32495; rank 25unknown | Original sourcewinner eligible | Third-party benchmark |
| #5 | GLM-5.1Z.ai | 1463.00 | — | 44,023 | glm-5.1; 95% CI [1459.00038238, 1467.00730085]; votes 44023; rank 30unknown | Original sourcewinner eligible | Third-party benchmark |
| #6 | Kimi-K2.6Moonshot AI | 1455.12 | — | 37,545 | kimi-k2.6; 95% CI [1450.59563552, 1459.63772982]; votes 37545; rank 34unknown | Original sourcewinner eligible | Third-party benchmark |
| #7 | GPT-5.6 SolOpenAI | 1454.12 | — | 21,537 | gpt-5.6-sol-xhigh; 95% CI [1448.87956591, 1459.35696358]; votes 21537; rank 36unknown | Original sourcewinner eligible | Third-party benchmark |
| #8 | DeepSeek-V4-ProDeepSeek | 1451.05 | — | 54,214 | deepseek-v4-pro; 95% CI [1447.05646256, 1455.03822989]; votes 54214; rank 40unknown | Original sourcewinner eligible | Third-party benchmark |
| #9 | GPT-5.6 TerraOpenAI | 1446.86 | — | 22,345 | gpt-5.6-terra-xhigh; 95% CI [1441.73650778, 1451.98969869]; votes 22345; rank 47unknown | Original sourcewinner eligible | Third-party benchmark |
| #10 | GLM-5Z.ai | 1446.36 | — | 27,649 | glm-5; 95% CI [1442.02946298, 1450.68440406]; votes 27649; rank 49unknown | Original sourcewinner eligible | Third-party benchmark |
| #11 | Kimi-K2.5Moonshot AI | 1445.68 | — | 70,610 | kimi-k2.5-thinking; 95% CI [1442.33624938, 1449.02096442]; votes 70610; rank 50unknown | Original sourcewinner eligible | Third-party benchmark |
| #12 | Claude Sonnet 5Anthropic | 1442.30 | — | 29,477 | claude-sonnet-5-high; 95% CI [1437.50389026, 1447.10454888]; votes 29477; rank 58unknown | Original sourcewinner eligible | Third-party benchmark |
| #13 | Hy3Tencent | 1440.53 | — | 6,466 | hy3; 95% CI [1432.70469713, 1448.35220311]; votes 6466; rank 62unknown | Original sourcewinner eligible | Third-party benchmark |
| #14 | Qwen3.5-397B-A17BQwen | 1438.08 | — | 71,882 | qwen3.5-397b-a17b; 95% CI [1434.68975454, 1441.46559151]; votes 71882; rank 68unknown | Original sourcewinner eligible | Third-party benchmark |
| #15 | Qwen3.8-27BQwen | 1436.96 | — | 5,046 | qwen3.8-27b; 95% CI [1428.42636059, 1445.49436282]; votes 5046; rank 73unknown | Original sourcewinner eligible | Third-party benchmark |
| #16 | MiniMax-M3MiniMax | 1434.24 | — | 42,988 | minimax-m3; 95% CI [1429.85087552, 1438.63891992]; votes 42988; rank 78unknown | Original sourcewinner eligible | Third-party benchmark |
| #17 | DeepSeek-V4-FlashDeepSeek | 1431.86 | — | 48,947 | deepseek-v4-flash; 95% CI [1427.72611296, 1435.99831462]; votes 48947; rank 80unknown | Original sourcewinner eligible | Third-party benchmark |
| #18 | GPT-5.6 LunaOpenAI | 1429.29 | — | 22,852 | gpt-5.6-luna-xhigh; 95% CI [1424.18984000, 1434.39657706]; votes 22852; rank 83unknown | Original sourcewinner eligible | Third-party benchmark |
| #19 | DeepSeek-V3.2DeepSeek | 1424.85 | — | 46,493 | deepseek-v3.2; 95% CI [1421.32068535, 1428.38405594]; votes 46493; rank 90unknown | Original sourcewinner eligible | Third-party benchmark |
| #20 | Mistral-Medium-3.5-128BMistral AI | 1421.26 | — | 11,006 | mistral-medium-3.5; 95% CI [1414.71769986, 1427.80720030]; votes 11006; rank 98unknown | Original sourcewinner eligible | Third-party benchmark |
| #21 | MiniMax-M2.7MiniMax | 1405.64 | — | 63,907 | minimax-m2.7; 95% CI [1401.94068888, 1409.34168102]; votes 63907; rank 133unknown | Original sourcewinner eligible | Third-party benchmark |
| #22 | Qwen3.5-35B-A3BQwen | 1395.76 | — | 29,059 | qwen3.5-35b-a3b; 95% CI [1391.45555487, 1400.05681685]; votes 29059; rank 143unknown | Original sourcewinner eligible | Third-party benchmark |
| #23 | Claude Haiku 4.5Anthropic | 1395.09 | — | 123.3K | claude-haiku-4-5-20251001; 95% CI [1392.53354833, 1397.64754357]; votes 123321; rank 144unknown | Original sourcewinner eligible | Third-party benchmark |
| #24 | DeepSeek-R1DeepSeek | 1372.49 | — | 18,524 | deepseek-r1; 95% CI [1367.64968138, 1377.33535654]; votes 18524; rank 164unknown | Original sourcewinner eligible | Third-party benchmark |
| #25 | gpt-oss-120bOpenAI | 1365.92 | — | 29,938 | gpt-oss-120b; 95% CI [1361.55097111, 1370.29626132]; votes 29938; rank 172unknown | Original sourcewinner eligible | Third-party benchmark |
| #26 | MiniMax-M2.5MiniMax | 1359.09 | — | 40,887 | minimax-m2.5; 95% CI [1355.12314583, 1363.05641474]; votes 40887; rank 179unknown | Original sourcewinner eligible | Third-party benchmark |
| #27 | NVIDIA-Nemotron-3-Nano-30B-A3B-BF16NVIDIA | 1348.33 | — | 15,332 | nvidia-nemotron-3-nano-30b-a3b-bf16; 95% CI [1342.87010235, 1353.78054224]; votes 15332; rank 190unknown | Original sourcewinner eligible | Third-party benchmark |
| #28 | DeepSeek-V3DeepSeek | 1332.29 | — | 21,770 | deepseek-v3; 95% CI [1327.56053953, 1337.02226188]; votes 21770; rank 205unknown | Original sourcewinner eligible | Third-party benchmark |
| #29 | Olmo-3.1-32B-InstructAi2 | 1311.93 | — | 11,463 | olmo-3.1-32b-instruct; 95% CI [1305.76029426, 1318.09944641]; votes 11463; rank 221unknown | Original sourcewinner eligible | Third-party benchmark |
| #30 | Olmo-3-32B-ThinkAi2 | 1299.59 | — | 5,682 | olmo-3-32b-think; 95% CI [1291.27927286, 1307.89702196]; votes 5682; rank 228unknown | Original sourcewinner eligible | Third-party benchmark |
| #31 | granite-4.1-8bIBM | 1291.92 | — | 4,032 | granite-4.1-8b; 95% CI [1281.83897318, 1302.00717592]; votes 4032; rank 234unknown | Original sourcewinner eligible | Third-party benchmark |
| #32 | gpt-oss-20bOpenAI | 1287.24 | — | 10,385 | gpt-oss-20b; 95% CI [1280.83748126, 1293.63325353]; votes 10385; rank 239unknown | Original sourcewinner eligible | Third-party benchmark |
| #33 | Llama-4-Maverick-17B-128E-InstructMeta | 1287.09 | — | 39,347 | llama-4-maverick-17b-128e-instruct; 95% CI [1282.82796557, 1291.34252783]; votes 39347; rank 240unknown | Original sourcewinner eligible | Third-party benchmark |
| #34 | Llama-4-Scout-17B-16E-InstructMeta | 1279.20 | — | 29,740 | llama-4-scout-17b-16e-instruct; 95% CI [1274.49093519, 1283.91406293]; votes 29740; rank 252unknown | Original sourcewinner eligible | Third-party benchmark |
| #35 | Llama-3.3-70B-InstructMeta | 1273.93 | — | 54,416 | llama-3.3-70b-instruct; 95% CI [1270.45747418, 1277.41238757]; votes 54416; rank 255unknown | Original sourcewinner eligible | Third-party benchmark |
| #36 | Llama-3.1-70B-InstructMeta | 1260.71 | — | 55,240 | llama-3.1-70b-instruct; 95% CI [1256.99798581, 1264.42325848]; votes 55240; rank 269unknown | Original sourcewinner eligible | Third-party benchmark |
| #37 | phi-4Microsoft | 1216.50 | — | 24,126 | phi-4; 95% CI [1211.91861833, 1221.07481125]; votes 24126; rank 293unknown | Original sourcewinner eligible | Third-party benchmark |
| #38 | Llama-3.1-8B-InstructMeta | 1186.38 | — | 49,605 | llama-3.1-8b-instruct; 95% CI [1182.28064178, 1190.48111360]; votes 49605; rank 306unknown | Original sourcewinner eligible | Third-party benchmark |
38 models ranked by the best winner-eligible current-version score when available, otherwise the best publisher-artifact score; highest first. Observed Aug 29, 2026.
Ranking protocol
arena rating · rating. Each model appears once; verified winner-eligible runs take precedence, and missing models are not imputed.
Acquisition coverage
38 published models from 395 source rows; 357 source identities remain quarantined.
Primary evidence