LMArena · Overall
We regularly fetch and display public benchmark data: LMArena user voting rankings, Epoch AI test scores, and OpenRouter usage rankings. Each leaderboard shows both its publication date and the date we retrieved it. These scores are not assigned by this site.
How to read this leaderboard
What it measures: This LMArena ranking aggregates user votes comparing responses from two models whose names are hidden. Like the LMArena website's default, it uses 'style and length adjustment (style control)' rankings that statistically adjust for the influence of response length and formatting, such as Markdown, on votes.
How to read the score: A relative score estimated from votes using the Bradley–Terry model and displayed on an Elo-like scale. The 95% confidence intervals indicate estimation uncertainty. Overlapping intervals alone do not establish equal capability.
Caveats: Because it evaluates which answers users prefer, results can be influenced by tone, length and formatting as well as accuracy. Also check the applied filters and adjustment settings.
Insights from this ranking
- The 95% confidence intervals for 1st-place gemini-4-argon-high and 2nd-place claude-opus-4-6-high do not overlap. This aggregation provides relatively clear evidence that the 1st-place model leads.
- Anthropic has the most models among the top 10, with 7.
- The 1st-place model received 4,932 votes, and this leaderboard includes 413 models.
Overall LMArena
Published 2026-10-02 · Retrieved 2026-10-04
Dots show scores; horizontal lines show 95% confidence intervals. Overlapping intervals suggest similar performance. The chart shows only the highest-scoring reasoning setting, such as high or max, for each model. The table shows rankings for all settings. Chat and vision rankings use 'style and length adjustment (style control)', the same default as the LMArena website.
Top ranking by company
- Googlegemini-4-argon-high#1
- Anthropicclaude-opus-4-6-high#2
- Metamuse-spark-1.3-max#9
- Moonshot AIkimi-k3-max#16
- OOpenAIgpt-5.6-sol-xhigh#20
- Alibabaqwen3.8-max#23
- Xiaomimimo-v2.6-pro#26
- ZZ.aiglm-5.3-max#27
- xxAIgrok-4.20-beta1#36
View table (top 50)
| Rank | Model | Developer | Score | 95% confidence interval | Votes |
|---|---|---|---|---|---|
| 1 | #1gemini-4-argon-high | 1525 | 1516–1534 | 4,932 | |
| 2 | #2claude-opus-4-6-high | Anthropic | 1505 | 1501–1508 | 77,636 |
| 3 | #3claude-fable-5-high | Anthropic | 1504 | 1500–1509 | 38,387 |
| 4 | #4claude-opus-5.5-high | Anthropic | 1504 | 1495–1513 | 4,552 |
| 5 | #5claude-opus-4-7-high | Anthropic | 1501 | 1498–1505 | 64,946 |
| 6 | #6claude-fable-5.1-max | Anthropic | 1501 | 1495–1508 | 11,800 |
| 7 | #7claude-opus-4-6 | Anthropic | 1497 | 1494–1501 | 82,189 |
| 8 | #8gemini-3.8-flash-high | 1495 | 1490–1500 | 26,298 | |
| 9 | #9muse-spark-1.3-max | Meta | 1494 | 1488–1500 | 12,343 |
| 10 | #10claude-opus-4-7 | Anthropic | 1494 | 1490–1498 | 66,058 |
| 11 | #11muse-spark-1.2 (xHigh) | Meta | 1494 | 1484–1503 | 4,356 |
| 12 | #12muse-spark-1.1 | Meta | 1491 | 1487–1496 | 37,238 |
| 13 | #13claude-opus-5-high | Anthropic | 1490 | 1486–1494 | 59,482 |
| 14 | #14claude-opus-5-max | Anthropic | 1489 | 1485–1494 | 28,937 |
| 15 | #15muse-spark | Meta | 1489 | 1484–1495 | 14,120 |
| 16 | #16kimi-k3-max | Moonshot AI | 1488 | 1484–1493 | 28,280 |
| 17 | #17gemini-3.7-flash-high | 1488 | 1483–1493 | 21,539 | |
| 18 | #18gemini-3.1-pro-preview | 1487 | 1484–1490 | 121,806 | |
| 19 | #19gemini-3-pro | 1485 | 1482–1489 | 41,910 | |
| 20 | #20Ogpt-5.6-sol-xhigh | OpenAI | 1484 | 1479–1488 | 36,656 |
| 21 | #21Ogpt-6.1-sol-max | OpenAI | 1483 | 1473–1494 | 3,071 |
| 22 | #22gemini-3.6-flash-high | 1483 | 1479–1487 | 36,070 | |
| 23 | #23qwen3.8-max | Alibaba | 1482 | 1476–1487 | 23,353 |
| 24 | #24claude-opus-4-8-high | Anthropic | 1482 | 1478–1485 | 63,974 |
| 25 | #25Ogpt-5.5-high | OpenAI | 1481 | 1478–1485 | 69,202 |
| 26 | #26mimo-v2.6-pro | Xiaomi | 1480 | 1471–1489 | 4,056 |
| 27 | #27Zglm-5.3-max | Z.ai | 1478 | 1473–1484 | 17,857 |
| 28 | #28gemini-3.5-flash-high | 1477 | 1473–1481 | 48,420 | |
| 29 | #29Ogpt-6-astra-max | OpenAI | 1477 | 1470–1484 | 9,156 |
| 30 | #30Ogpt-5.5 | OpenAI | 1477 | 1473–1480 | 70,781 |
| 31 | #31gemini-3.5-flash-medium | 1476 | 1472–1480 | 47,187 | |
| 32 | #32Zglm-5.2-max | Z.ai | 1476 | 1471–1480 | 45,399 |
| 33 | #33Ogpt-5.2-chat-latest-20260210 | OpenAI | 1476 | 1472–1480 | 35,937 |
| 34 | #34Ogpt-5.4-high | OpenAI | 1475 | 1471–1479 | 64,642 |
| 35 | #35qwen3.7-max-preview | Alibaba | 1475 | 1465–1485 | 3,920 |
| 36 | #36xgrok-4.20-beta1 | xAI | 1475 | 1470–1479 | 27,827 |
| 37 | #37claude-opus-4-8 | Anthropic | 1475 | 1471–1478 | 65,072 |
| 38 | #38deepseek-v4.1-flash-max | DeepSeek | 1474 | 1468–1481 | 8,728 |
| 39 | #39claude-opus-4-5-20251101-high-32k | Anthropic | 1474 | 1470–1477 | 37,448 |
| 40 | #40Ogpt-5.5-instant | OpenAI | 1473 | 1468–1478 | 27,098 |
| 41 | #41Zglm-5.3-flash | Z.ai | 1473 | 1468–1478 | 22,971 |
| 42 | #42gemini-3-flash | 1473 | 1468–1477 | 31,259 | |
| 43 | #43claude-sonnet-4-6 | Anthropic | 1472 | 1469–1476 | 70,747 |
| 44 | #44xgrok-4.20-beta-0309-reasoning | xAI | 1472 | 1468–1475 | 66,334 |
| 45 | #45claude-sonnet-5.5-xhigh | Anthropic | 1471 | 1461–1481 | 3,145 |
| 46 | #46xgrok-4.20-multi-agent-beta-0309 | xAI | 1471 | 1467–1474 | 64,788 |
| 47 | #47claude-opus-4-5-20251101 | Anthropic | 1470 | 1467–1473 | 72,772 |
| 48 | #48ernie-5.1 | Baidu | 1468 | 1463–1472 | 39,396 |
| 49 | #49mimo-v2.5-pro | Xiaomi | 1468 | 1464–1471 | 70,871 |
| 50 | #50xgrok-4.5 | xAI | 1466 | 1461–1470 | 39,789 |
Data: LMArena leaderboard dataset (CC BY 4.0). Changes: Top 50 entries per leaderboard; scores are rounded. https://huggingface.co/datasets/lmarena-ai/leaderboard-datasetCompany logos are trademarks of their respective owners and are used only for identification (icons: Simple Icons).Data: Epoch AI — Capabilities & benchmarking (CC BY 4.0). https://epoch.ai/benchmarksSource: OpenRouter (openrouter.ai/rankings), as of 2026-10-04. Licensed under CC BY 4.0. https://openrouter.ai/rankings