A style-adjusted Arena ranking that regresses out response length and markdown formatting so ratings lean more on substance than presentation.
unassessed
| Category | human-preference |
|---|---|
| Subcategory | pairwise human preference, style-adjusted |
| Page status | active |
| Metric | Elo |
| Direction | higher_is_better |
| Publisher | LMSYS (Large Model Systems Organization), UC Berkeley Sky Computing Lab (founding org); operates today as Arena (formerly LMArena) |
arena_elo_style_control applies a style adjustment to the same votes and the same Bradley-Terry fit used elsewhere in the arena_elo family, rather than drawing on a different vote pool. LMSYS adds response length and markdown formatting (header, bold and list counts, each expressed as a normalised difference between the two compared responses) as extra regressors in the logistic regression that produces Arena ratings, so the resulting model coefficients are adjusted for — controlled for — those style effects instead of reflecting them. It can be layered onto Overall or onto a category such as Hard Prompts.
Same anonymous, randomized side-by-side text chat as the underlying category; ratings are recomputed with response-length and markdown-count features included in the regression.
| Model | Provider | Score | Card as of |
|---|---|---|---|
| Claude Opus 4.6 | Anthropic | 1502.8 | 2026-04 |
| Grok 4.20 (Reasoning) | xAI | 1490.4 | 2026-04 |
| GPT-5.4 | OpenAI | 1483.6 | 2026-04 |
| GPT-5.2 | OpenAI | 1477.0 | 2026-04 |
| Grok 4.20 Multi-Agent | xAI | 1474.5 | 2026-04 |
| Claude Opus 4.5 | Anthropic | 1473.1 | 2026-04 |
| Grok 4 | xAI | 1470.7 | 2026-04 |
| GLM 5.1 | Z.ai (Zhipu AI) | 1467.4 | 2026-04 |
| Claude Sonnet 4.6 | Anthropic | 1462.2 | 2026-04 |
| GPT-5.4 mini | OpenAI | 1457.8 | 2026-04 |
| GPT-5.3 Codex Spark | OpenAI | 1456.4 | 2026-04 |
| GLM 5 | Zhipu AI | 1455.6 | undated |
| GPT-5.1 | OpenAI | 1454.6 | 2026-04 |
| Claude Sonnet 4.5 | Anthropic | 1451.9 | 2026-04 |
| Gemma 4 31B | Google DeepMind | 1451.2 | 2026-04 |
| Claude Opus 4.1 | Anthropic | 1448.6 | 2026-04 |
| Gemini 2.5 Pro | Google DeepMind | 1448.2 | 2026-04 |
| GPT-4o | OpenAI | 1443.0 | 2026-04 |
| GLM 4.7 | Zhipu AI | 1442.7 | undated |
| gemma 4 26B A4B it | Google DeepMind | 1437.9 | 2026-04 |
| GPT-5 | OpenAI | 1433.4 | 2026-04 |
| o3 | OpenAI | 1431.2 | 2026-04 |
| Grok 4.1 Fast | xAI | 1431.1 | 2026-04 |
| GLM 4.6 | Zhipu AI | 1425.8 | undated |
| DeepSeek V3.2 | DeepSeek | 1424.4 | 2026-04 |
| Claude Opus 4 | Anthropic | 1423.7 | 2026-04 |
| DeepSeek V3.2 Exp | DeepSeek | 1422.8 | 2026-04 |
| DeepSeek R1 0528 | DeepSeek | 1421.7 | 2026-04 |
| Grok 4 Fast (Non-Reasoning) | xAI | 1420.7 | 2026-04 |
| DeepSeek V3.1 | DeepSeek | 1417.9 | 2026-04 |
| Qwen3 235B-A22B | Alibaba / Qwen Team | 1415.8 | 2026-04 |
| Mistral Large 3 | Mistral AI | 1415.0 | 2026-04 |
| GPT-4.1 | OpenAI | 1413.0 | 2026-04 |
| Grok 3 | xAI | 1411.7 | 2026-04 |
| Gemini 2.5 Flash | Google DeepMind | 1411.0 | 2026-04 |
| GLM 4.5 | Zhipu AI | 1410.9 | undated |
| Mistral Medium 3.1 | Mistral AI | 1410.3 | 2026-04 |
| Claude Haiku 4.5 | Anthropic | 1407.2 | 2026-04 |
| Gemini 2.5 Flash Preview 09-25 | Google DeepMind | 1405.1 | 2026-04 |
| GPT-5.4 nano | OpenAI | 1402.7 | 2026-04 |
| o1 | OpenAI | 1401.5 | 2026-04 |
| Claude Sonnet 4 | Anthropic | 1398.5 | 2026-04 |
| DeepSeek R1 | DeepSeek | 1397.5 | 2026-04 |
| DeepSeek V3 0324 | DeepSeek | 1394.6 | 2026-04 |
| o4-mini | OpenAI | 1389.7 | 2026-04 |
| GPT-5 Mini | OpenAI | 1389.5 | 2026-04 |
| o1-preview | OpenAI | 1387.6 | 2026-04 |
| Qwen3-Coder 480B-A35B Instruct | Alibaba / Qwen Team | 1387.4 | 2026-04 |
| Claude Sonnet 3.7 | Anthropic | 1386.3 | 2026-04 |
| Mistral Medium 3 | Mistral AI | 1386.2 | 2026-04 |
| GPT-4.1 mini | OpenAI | 1382.1 | 2026-04 |
| Gemini 2.5 Flash Lite Preview 09-25 | Google DeepMind | 1380.0 | 2026-04 |
| GLM 4.6V | Zhipu AI | 1377.9 | undated |
| Gemini 2.5 Flash Lite | Google DeepMind | 1374.3 | 2026-04 |
| GLM 4.5 Air | Zhipu AI | 1372.8 | undated |
| Claude Sonnet 3.5 v2 | Anthropic | 1371.4 | 2026-04 |
| GLM 4.7 Flash | Zhipu AI | 1368.7 | undated |
| Gemma 3 27B | Google DeepMind | 1365.1 | 2026-04 |
| o3-mini | OpenAI | 1363.2 | 2026-04 |
| Grok 3 Mini | xAI | 1362.7 | 2026-04 |
| Gemini 2.0 Flash | Google DeepMind | 1360.0 | 2026-04 |
| DeepSeek V3 | DeepSeek | 1358.2 | 2026-04 |
| Mistral Small 3.2 | Mistral AI | 1357.0 | 2026-04 |
| Command A | Cohere | 1353.4 | 2026-04 |
| GLM 4.5V | Zhipu AI | 1353.3 | undated |
| Gemini 2.0 Flash Lite | Google DeepMind | 1352.8 | 2026-04 |
| Gemini 1.5 Pro | Google DeepMind | 1350.6 | 2026-04 |
| Qwen3 32B | Alibaba / Qwen Team | 1347.0 | 2026-04 |
| GPT-4o (2024-05-13) | OpenAI | 1345.1 | 2026-04 |
| Gemma 3 12B | Google DeepMind | 1341.4 | 2026-04 |
| Claude Sonnet 3.5 | Anthropic | 1341.3 | 2026-04 |
| o1-mini | OpenAI | 1336.6 | 2026-04 |
| GPT-5 Nano | OpenAI | 1336.4 | 2026-04 |
| Grok 2 | xAI | 1334.8 | 2026-04 |
| GPT-4o (2024-08-06) | OpenAI | 1334.3 | 2026-04 |
| Llama 3.1 405B Instruct | Meta | 1334.2 | 2026-04 |
| Llama 3.1 405B Instruct FP8 | Meta | 1332.4 | 2026-04 |
| Qwen3 30B-A3B | Alibaba / Qwen Team | 1327.3 | 2026-04 |
| Llama 4 Maverick 17B 128E Instruct | Meta | 1326.6 | 2026-04 |
| GPT-4 Turbo | OpenAI | 1323.4 | 2026-04 |
| Claude Haiku 3.5 | Anthropic | 1322.4 | 2026-04 |
| Llama 4 Scout 17B 16E Instruct | Meta | 1321.9 | 2026-04 |
| GPT-4.1 nano | OpenAI | 1321.4 | 2026-04 |
| Claude Opus 3 | Anthropic | 1320.7 | 2026-04 |
| Llama-3.3-70B-Instruct | Meta | 1318.0 | 2026-04 |
| GPT-4o mini | OpenAI | 1317.2 | 2026-04 |
| Gemini 1.5 Flash | Google DeepMind | 1309.1 | 2026-04 |
| Mistral Large 2.1 | Mistral AI | 1304.7 | 2026-04 |
| Magistral Medium (latest) | Mistral AI | 1303.0 | 2026-04 |
| Gemma 3 4B | Google DeepMind | 1302.8 | 2026-04 |
| Mistral Small 3.1 24B Instruct 2503 | Mistral AI | 1302.7 | 2026-04 |
| Qwen2.5 72B Instruct | Alibaba / Qwen Team | 1302.3 | 2026-04 |
| Llama 3.1 70B Instruct | Meta | 1292.8 | 2026-04 |
| gemma 2 27B it | Google DeepMind | 1287.6 | 2026-04 |
| Claude Sonnet 3 | Anthropic | 1279.8 | 2026-04 |
| Command R+ | Cohere | 1275.5 | 2026-04 |
| Mistral Small 24B Instruct 2501 | Mistral AI | 1273.5 | 2026-04 |
| gemma 2 9B it | Google DeepMind | 1265.0 | 2026-04 |
| Claude Haiku 3 | Anthropic | 1259.9 | 2026-04 |
| Gemini 1.5 Flash-8B | Google DeepMind | 1258.1 | 2026-04 |
| phi 4 | Microsoft | 1255.4 | 2026-04 |
| Command R | Cohere | 1249.1 | 2026-04 |
| Llama 3.1 8B Instruct | Meta | 1211.0 | 2026-04 |