Benchmarks · composite
Arena Text · German
Crowdsourced pairwise human preference on open-ended text prompts (style control on); Bradley-Terry rating. Only the German category.
187 models · 198 results · data as of 2026-10-29 · arena.ai ↗
Leaderboard
Gemini 4 Argon high
All results
198 rows| # | Model | Effort | Arena Text · German |
|---|---|---|---|
| 1 | Gemini 4 Argon | high | 1550 ±44 |
| 2 | Gemini 3 Pro | high | 1521 ±22 |
| 3 | Muse Spark | 1512 ±40 | |
| 4 | Gemini 3.7 Flash | high | 1510 ±27 |
| 5 | Opus 5 | max | 1508 ±25 |
| 6 | Opus 4.6 | high | 1507 ±17 |
| 7 | Muse Spark 1.3 | max | 1507 ±32 |
| 8 | Opus 4.7 | high | 1506 ±19 |
| 9 | Kimi K3 | max | 1505 ±23 |
| 10 | GLM 5.3 | max | 1504 ±29 |
| 11 | GPT-5.6 Sol | xhigh | 1504 ±23 |
| 12 | Gemini 3.8 Flash | high | 1500 ±22 |
| 13 | Gemini 3 Flash | high | 1499 ±25 |
| 14 | Fable 5 | high | 1498 ±23 |
| 15 | Opus 4.7 | max | 1497 ±19 |
| 16 | Muse Spark 1.2 | xhigh | 1496 ±43 |
| 17 | Opus 4.6 | max | 1494 ±17 |
| 18 | Gemini 3.1 Pro | high | 1494 ±14 |
| 19 | Opus 5.5 | high | 1492 ±45 |
| 20 | GPT-6 Astra | max | 1492 ±39 |
| 21 | Opus 4.8 | high | 1491 ±19 |
| 22 | GPT-5.5 | xhigh | 1491 ±19 |
| 23 | Gemini 3.5 Flash | medium | 1491 ±23 |
| 24 | DeepSeek V4.1 Flash | max | 1489 ±36 |
| 25 | Grok 4.20 | xhigh | 1489 ±28 |
| 26 | Fable 5.1 | max | 1486 ±33 |
| 27 | Opus 5 | high | 1486 ±19 |
| 28 | Gemini 3.5 Flash | high | 1486 ±22 |
| 29 | Gemini 3.6 Flash | high | 1481 ±24 |
| 30 | GPT-5.2 | xhigh | 1481 ±25 |