evals.report
BenchmarksLabsCompareRun guidesIn the wild
BenchmarksChat preference

LMArena

A public chat-preference evaluation surface with source-defined preference ratings and model comparisons.

Chat preferencesource-defined ratingHigher is better
ScoresAboutRun this benchmark

What is LMArena?

A public chat-preference evaluation surface with source-defined preference ratings and model comparisons. evals.report tracks reported LMArena scores with the model, source, status, date, and run caveats attached — official leaderboard scores, vendor-reported launches, and clearly labeled community runs.

Top reported LMArena score: Claude Opus 4.6 — 1499 (source-defined rating).

ModelLabScore↓Source modelStatusDate
Claude Opus 4.6Anthropic1499claude-opus-4-6-thinkingOfficialFeb 5, 2026Details
Claude Opus 4.7Anthropic1486claude-opus-4-7-thinkingOfficialApr 16, 2026Details
Gemini 3.5 FlashGoogle DeepMind1482gemini-3.5-flashOfficialMay 19, 2026Details
Gemini 3.1 Pro PreviewGoogle DeepMind1481gemini-3.1-pro-previewOfficialFeb 19, 2026Details
Gemini 3 ProGoogle DeepMind1479gemini-3-proOfficialNov 18, 2025Details
Qwen3.7 Max PreviewAlibaba / Qwen1474qwen3.7-max-previewOfficialMay 14, 2026Details
Muse SparkMeta1474muse-sparkOfficialApr 8, 2026Details
GPT-5.4OpenAI1472gpt-5.4-highOfficialMar 5, 2026Details
Qwen3.5 Max PreviewAlibaba / Qwen1470qwen3.5-max-previewOfficialMar 25, 2026Details
ERNIE 5.1Baidu1469ernie-5.1OfficialMay 8, 2026Details
GLM-5.1OpenZ.ai1469glm-5.1OfficialApr 7, 2026Details
GPT-5.5OpenAI1468gpt-5.5-highOfficialApr 23, 2026Details
Gemini 3 FlashGoogle DeepMind1466gemini-3-flashOfficialDec 17, 2025Details
Gemini 2.5 ProGoogle DeepMind1457gemini-2.5-proOfficialMar 25, 2025Details
Kimi K2.6OpenMoonshot AI1456kimi-k2.6OfficialApr 20, 2026Details
Claude Sonnet 4.6Anthropic1454claude-sonnet-4-6OfficialFeb 17, 2026Details
Grok 4.20 beta reasoningxAI1453grok-4.20-beta-0309-reasoningOfficialMar 9, 2026Details
Claude Opus 4.5Anthropic1449claude-opus-4-5-20251101OfficialNov 24, 2025Details
DeepSeek V4 ProOpenDeepSeek1446deepseek-v4-proOfficialApr 24, 2026Details
GLM-5OpenZ.ai1445glm-5OfficialFeb 11, 2026Details
Qwen 3.6 Max PreviewAlibaba / Qwen1444qwen3.6-max-previewOfficialApr 20, 2026Details
GLM-4.6OpenZ.ai1440glm-4.6OfficialSep 30, 2025Details
Qwen 3.6 PlusAlibaba / Qwen1437qwen3.6-plusOfficialApr 2, 2026Details
Claude Sonnet 4.5Anthropic1437claude-sonnet-4-5-20250929OfficialSep 29, 2025Details
DeepSeek V3.2OpenDeepSeek1423deepseek-v3.2OfficialDec 1, 2025Details
GPT-5.1OpenAI1422gpt-5.1OfficialNov 12, 2025Details
Qwen3 235B A22B Instruct 2507OpenAlibaba / Qwen1419qwen3-235b-a22b-instruct-2507OfficialJul 21, 2025Details
Gemini 2.5 FlashGoogle DeepMind1417gemini-2.5-flashOfficialApr 17, 2025Details
GPT-5.2OpenAI1411gpt-5.2OfficialDec 11, 2025Details
Grok 4xAI1409grok-4-0709OfficialJul 9, 2025Details
Grok 4.3xAI1408grok-4.3OfficialApr 17, 2026Details
GPT-5OpenAI1405gpt-5-highOfficialAug 7, 2025Details
DeepSeek V3 0324OpenDeepSeek1374deepseek-v3-0324OfficialMar 24, 2025Details
DeepSeek R1OpenDeepSeek1372deepseek-r1OfficialJan 20, 2025Details
GPT-OSS-120BOpenOpenAI1365gpt-oss-120bOfficialAug 5, 2025Details
Claude Opus 4Anthropic1364claude-opus-4-20250514OfficialMay 22, 2025Details
Claude Sonnet 4Anthropic1337claude-sonnet-4-20250514OfficialMay 22, 2025Details
DeepSeek V3OpenDeepSeek1332deepseek-v3OfficialDec 26, 2024Details
Claude 3.7 SonnetAnthropic1299claude-3-7-sonnet-20250219OfficialFeb 24, 2025Details
Claude 3.5 SonnetAnthropic1297claude-3-5-sonnet-20241022OfficialJun 20, 2024Details
Mistral LargeMistral AI1265mistral-large-2411OfficialFeb 26, 2024Details

Each row reports the model’s source-defined rating on LMArena. Click a row for the full run context.