Arena Text
Overall text arena rating with style control.
As of 24 Oct 2026, Gemini 4 Argon leads Arena Text on BenchLeader with 1517, ahead of Claude Opus 5.5 at 1509, across 373 model configurations with a published result.
- Published by
- Arena (LMArena)
- Category
- Human preference
- Index weight
- 1.0
- Models
- 373
- Data as of
- 24 Oct 2026
- Newest result
- 22 Oct 2026
CC BY 4.0 — lmarena-ai/leaderboard-dataset on Hugging Face.
What the test looks like
LMArena shows two anonymous models the same prompt and asks the visitor which answer is better. Millions of such votes are combined into a rating, here with style control that discounts the effect of length and formatting.
How it is scored
A Bradley-Terry rating in Elo-like units: a 100-point gap means the higher model wins about 64% of head-to-heads. Reported with a confidence interval.
What to keep in mind
It measures what people prefer in chat, which favours agreeable, well-formatted answers, and the voter population is self-selected. Providers can test unreleased variants anonymously, and votes on newly listed models take weeks to settle.
- 1Gemini 4 Argon (high)1517
- 2Claude Opus 5.5 (high)1509
- 3Muse Wonder (max)1509
- 4Claude Fable 5 (high)1503
- 5Claude Fable 5.1 (max)1503
- 6Claude Opus 4.6 (high)1503
- 7Claude Opus 4.7 (high)1501
- 8Claude Opus 4.61497
- 9Gemini 3.8 Flash (high)1496
- 10Muse Spark 1.3 (max)1496
- 11Claude Opus 4.71494
- 12Muse Spark 1.2 (xhigh)1491
- 13Claude Opus 5 (max)1491
- 14Claude Opus 5 (high)1491
- 15Muse Spark 1.11490
| # | ||||
|---|---|---|---|---|
| 1 |  Gemini 4 Argonhigh | 1517 | 70.8 | 2026-10-22 |
| 2 |  Claude Opus 5.5high | 1509 | 69.4 | 2026-10-22 |
| 3 |  Muse Wondermax | 1509 | 67.5 | 2026-10-22 |
| 4 | Claude Fable 5high | 1503 | 64.6 | 2026-10-22 |
| 5 | Claude Fable 5.1max | 1503 | 70.5 | 2026-10-22 |
| 6 | Claude Opus 4.6high | 1503 | 62.7 | 2026-10-22 |
| 7 | Claude Opus 4.7high | 1501 | 61.0 | 2026-10-22 |
| 8 | Claude Opus 4.6 | 1497 | 62.6 | 2026-10-22 |
| 9 | Gemini 3.8 Flashhigh | 1496 | 64.4 | 2026-10-22 |
| 10 | Muse Spark 1.3max | 1496 | 67.1 | 2026-10-22 |
| 11 | Claude Opus 4.7 | 1494 | 63.3 | 2026-10-22 |
| 12 | Muse Spark 1.2xhigh | 1491 | 63.2 | 2026-10-22 |
| 13 | Claude Opus 5max | 1491 | 68.8 | 2026-10-22 |
| 14 | Claude Opus 5high | 1491 | 67.9 | 2026-10-22 |
| 15 | Muse Spark 1.1 | 1490 | 62.5 | 2026-10-22 |
| 16 | Gemini 3.7 Flashhigh | 1490 | 63.5 | 2026-10-22 |
| 17 | Muse Spark | 1489 | 64.3 | 2026-10-22 |
| 18 |  Kimi K3maxopen ↗ | 1487 | 64.7 | 2026-10-22 |
| 19 | Gemini 3.1 Pro | 1487 | 60.6 | 2026-10-22 |
| 20 | Gemini 3 Pro | 1486 | 60.4 | 2026-10-22 |
| 21 |  GPT-5.6 Solxhigh | 1485 | 66.6 | 2026-10-22 |
| 22 | Gemini 3.6 Flashhigh | 1483 | 62.0 | 2026-10-22 |
| 23 |  Qwen3.8 Maxmax | 1483 | 63.3 | 2026-10-22 |
| 24 | GPT-5.5high | 1482 | 65.3 | 2026-10-22 |
| 25 | Claude Opus 4.8high | 1482 | 60.7 | 2026-10-22 |
| 26 |  GLM 5.3maxopen ↗ | 1481 | 63.5 | 2026-10-22 |
| 27 | GPT-6.1 Solmax | 1480 | 69.9 | 2026-10-22 |
| 28 |  MiMo-V2.6-Proopen ↗ | 1480 | 64.0 | 2026-10-22 |
| 29 | Claude Sonnet 5.5xhigh | 1478 | 67.9 | 2026-10-22 |
| 30 | Gemini 3.5 Flashhigh | 1477 | 63.1 | 2026-10-22 |
| 31 | GPT-5.5 | 1477 | 60.3 | 2026-10-22 |
| 32 | GPT-6 Astramax | 1477 | 70.4 | 2026-10-22 |
| 33 | Gemini 3.5 Flashmedium | 1476 | 62.8 | 2026-10-22 |
| 34 | GLM 5.3 Flashopen ↗ | 1476 | 62.6 | 2026-10-22 |
| 35 |  DeepSeek V4.1 Flashmaxopen ↗ | 1476 | 61.2 | 2026-10-22 |
| 36 | GPT-5.2 | 1476 | 58.0 | 2026-10-22 |
| 37 | GLM 5.2maxopen ↗ | 1475 | 62.3 | 2026-10-22 |
| 38 | GPT-5.4high | 1475 | 57.9 | 2026-10-22 |
| 39 |  Grok 4.20 1 | 1475 | – | 2026-10-22 |
| 40 | Qwen3.7max | 1474 | 60.6 | 2026-10-22 |
| 41 | Claude Opus 4.8 | 1474 | 60.1 | 2026-10-22 |
| 42 | Claude Opus 4.5high | 1474 | 59.2 | 2026-10-22 |
| 43 | GPT-5.5 Instant | 1473 | 58.0 | 2026-10-22 |
| 44 | Gemini 3 Flash | 1473 | 57.0 | 2026-10-22 |
| 45 | Claude Sonnet 4.6 | 1472 | 57.1 | 2026-10-22 |
| 46 | Grok 4.20thinking | 1472 | 58.6 | 2026-10-22 |
| 47 | Grok 4.20 Multi-Agent | 1471 | 58.5 | 2026-10-22 |
| 48 | Claude Opus 4.5 | 1470 | 57.9 | 2026-10-22 |
| 49 | GPT-5.6 Terraxhigh | 1467 | 62.8 | 2026-10-22 |
| 50 |  ERNIE 5.1 | 1467 | – | 2026-10-22 |
| 51 | MiMo-V2.5-Proopen ↗ | 1467 | 54.9 | 2026-10-22 |
| 52 | Grok 4.5 | 1466 | 59.0 | 2026-10-22 |
| 53 | GLM 5.1open ↗ | 1465 | 55.9 | 2026-10-22 |
| 54 | Grok 4.1thinking | 1465 | 54.6 | 2026-10-22 |
| 55 | GPT-5.4 | 1465 | 58.2 | 2026-10-22 |
| 56 | Qwen3.5 Maxmax | 1465 | – | 2026-10-22 |
| 57 | DeepSeek V4 Prohighopen ↗ | 1465 | 57.5 | 2026-10-22 |
| 58 | Claude Sonnet 5high | 1462 | 60.2 | 2026-10-22 |
| 59 | Kimi K2.6open ↗ | 1461 | 55.4 | 2026-10-22 |
| 60 | Qwen3.6 Maxmax | 1461 | 62.3 | 2026-10-22 |
More human preference benchmarks · All LLM benchmarks
Cite as: BenchLeader, “Arena Text leaderboard”, data as of 24 Oct 2026, https://www.benchleader.com/benchmarks/lmarena_text.
Arena Text: questions
- What does Arena Text measure?
- LMArena shows two anonymous models the same prompt and asks the visitor which answer is better. Millions of such votes are combined into a rating, here with style control that discounts the effect of length and formatting. Scores are reported in Elo-style ratings from pairwise comparisons; higher is better.
- Which AI model leads Arena Text?
- Gemini 4 Argon leads Arena Text with 1517 as of 24 Oct 2026, ahead of Claude Opus 5.5 at 1509.
- How many models have Arena Text results?
- 373 model configurations have a Arena Text result on BenchLeader, all taken from Arena (LMArena).
- Who runs Arena Text and how often is it updated?
- Arena Text is published by Arena (LMArena). BenchLeader re-reads the published results once a day and records the date each result was published.
- Does Arena Text count toward the BenchLeader Index?
- Yes. Arena Text contributes to the human preference category of the BenchLeader Index, normalised so that 50 is the average of the evaluated models.