Benchmarks · composite
Arena Text · Russian
Crowdsourced pairwise human preference on open-ended text prompts (style control on); Bradley-Terry rating. Only the Russian category.
209 models · 220 results · updated 2026-11-09 · arena.ai ↗
Leaderboard
Gemini 4 Carbon high
All results
220 rows| # | Model | Effort | Arena Text · Russian |
|---|---|---|---|
| 1 | Gemini 4 Carbon | high | 1548 ±24 |
| 2 | Gemini 4 Argon | high | 1532 ±12 |
| 3 | Muse Wonder | max | 1522 ±15 |
| 4 | Gemini 3.8 Flash | high | 1517 ±8 |
| 5 | Gemini 3.7 Flash | high | 1515 ±10 |
| 6 | Fable 5 | high | 1514 ±9 |
| 7 | Opus 5.5 | high | 1512 ±13 |
| 8 | Fable 5.1 | max | 1507 ±12 |
| 9 | Opus 4.7 | high | 1505 ±8 |
| 10 | Opus 4.6 | high | 1505 ±7 |
| 11 | Opus 4.6 | max | 1504 ±7 |
| 12 | Gemini 3.1 Pro | high | 1501 ±6 |
| 13 | Gemini 3 Pro | high | 1500 ±9 |
| 14 | Muse Spark 1.3 | max | 1498 ±10 |
| 15 | GPT-5.6 Sol | xhigh | 1498 ±9 |
| 16 | Muse Spark 1.2 | xhigh | 1496 ±14 |
| 17 | Opus 5 | high | 1496 ±7 |
| 18 | Opus 4.8 | high | 1496 ±8 |
| 19 | Kimi K3 | max | 1494 ±9 |
| 20 | Muse Spark 1.1 | max | 1494 ±9 |
| 21 | Opus 4.7 | max | 1494 ±8 |
| 22 | Opus 5 | max | 1492 ±10 |
| 23 | Gemini 3.5 Flash | medium | 1492 ±9 |
| 24 | Gemini 3.6 Flash | high | 1492 ±10 |
| 25 | Gemini 3.5 Flash | high | 1491 ±9 |
| 26 | Qwen 3.8 Max | max | 1490 ±9 |
| 27 | GPT-6.1 Sol | max | 1487 ±14 |
| 28 | GPT-5.4 | high | 1487 ±8 |
| 29 | Opus 4.8 | max | 1486 ±8 |
| 30 | Gemini 3 Flash | high | 1486 ±9 |