Benchmarks · composite
Arena Text · Math
Crowdsourced pairwise human preference on open-ended text prompts (style control on); Bradley-Terry rating. Only the Math category.
202 models · 213 results · data as of 2026-10-29 · arena.ai ↗
Leaderboard
Gemini 4 Argon high
All results
213 rows| # | Model | Effort | Arena Text · Math |
|---|---|---|---|
| 1 | Gemini 4 Argon | high | 1532 ±22 |
| 2 | Fable 5.1 | max | 1528 ±20 |
| 3 | Gemini 4 Carbon | high | 1525 ±41 |
| 4 | Gemini 3.8 Flash | high | 1524 ±12 |
| 5 | Opus 5 | high | 1521 ±11 |
| 6 | Fable 5 | high | 1520 ±13 |
| 7 | Opus 4.6 | high | 1516 ±9 |
| 8 | Opus 5 | max | 1515 ±15 |
| 9 | Opus 5.5 | high | 1512 ±24 |
| 10 | GPT-6.1 Sol | max | 1509 ±25 |
| 11 | Muse Spark 1.3 | max | 1508 ±17 |
| 12 | Sonnet 5.5 | xhigh | 1508 ±26 |
| 13 | Opus 4.7 | high | 1504 ±11 |
| 14 | Gemini 3.6 Flash | high | 1503 ±14 |
| 15 | Opus 4.6 | max | 1503 ±9 |
| 16 | GLM 5.3 Flash | max | 1503 ±13 |
| 17 | Qwen 3.8 Max | max | 1503 ±15 |
| 18 | Gemini 3.7 Flash | high | 1503 ±15 |
| 19 | GPT-5.5 | xhigh | 1500 ±10 |
| 20 | GLM 5.3 | max | 1499 ±17 |
| 21 | Kimi K3 | max | 1499 ±15 |
| 22 | DeepSeek V4.1 Flash | max | 1498 ±21 |
| 23 | Gemini 3.5 Flash | high | 1497 ±12 |
| 24 | GPT-5.6 Sol | xhigh | 1495 ±13 |
| 25 | Muse Spark 1.1 | max | 1494 ±13 |
| 26 | GPT-5.5 | high | 1493 ±10 |
| 27 | Opus 4.8 | high | 1493 ±11 |
| 28 | Muse Spark 1.2 | xhigh | 1492 ±23 |
| 29 | Opus 4.7 | max | 1491 ±11 |
| 30 | GPT-5.4 | high | 1491 ±11 |