Benchmarks · composite
Arena Text
Crowdsourced pairwise human preference on open-ended text prompts (style control on); Bradley-Terry rating.
205 models · 216 results · data as of 2026-10-29 · arena.ai ↗
Leaderboard
Gemini 4 Carbon high
All results
216 rows| # | Model | Effort | Arena Text |
|---|---|---|---|
| 1 | Gemini 4 Carbon | high | 1528 ±9 |
| 2 | Gemini 4 Argon | high | 1520 ±6 |
| 3 | Opus 5.5 | high | 1509 ±7 |
| 4 | Fable 5.1 | max | 1504 ±6 |
| 5 | Fable 5 | high | 1504 ±4 |
| 6 | Opus 4.6 | high | 1503 ±3 |
| 7 | Opus 4.7 | high | 1501 ±4 |
| 8 | Opus 4.6 | max | 1496 ±3 |
| 9 | Gemini 3.8 Flash | high | 1496 ±4 |
| 10 | Muse Spark 1.3 | max | 1495 ±5 |
| 11 | Opus 4.7 | max | 1494 ±4 |
| 12 | Muse Spark 1.2 | xhigh | 1492 ±6 |
| 13 | Opus 5 | high | 1492 ±4 |
| 14 | Opus 5 | max | 1491 ±5 |
| 15 | Muse Spark 1.1 | max | 1490 ±4 |
| 16 | Gemini 3.7 Flash | high | 1490 ±5 |
| 17 | Muse Spark | 1489 ±6 | |
| 18 | Kimi K3 | max | 1488 ±4 |
| 19 | Gemini 3.1 Pro | high | 1487 ±3 |
| 20 | Gemini 3 Pro | high | 1485 ±4 |
| 21 | GPT-5.6 Sol | xhigh | 1485 ±4 |
| 22 | Gemini 3.6 Flash | high | 1483 ±4 |
| 23 | Qwen 3.8 Max | max | 1482 ±5 |
| 24 | Opus 4.8 | high | 1482 ±4 |
| 25 | GPT-5.5 | high | 1482 ±4 |
| 26 | GPT-6.1 Sol | max | 1481 ±7 |
| 27 | GLM 5.3 | max | 1481 ±5 |
| 28 | MiMo V2.6 Pro | high | 1480 ±9 |
| 29 | Gemini 3.5 Flash | high | 1478 ±4 |
| 30 | Sonnet 5.5 | xhigh | 1477 ±7 |