Benchmarks · composite
Arena Text · Industry: software and IT
Crowdsourced pairwise human preference on open-ended text prompts (style control on); Bradley-Terry rating. Only the Industry: software and IT category.
212 models · 223 results · updated 2026-11-09 · arena.ai ↗
Leaderboard
Gemini 4 Carbon high
All results
223 rows| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Gemini 4 Carbon | high | 1550 ±15 |
| 2 | Gemini 4 Argon | high | 1549 ±8 |
| 3 | Muse Wonder | max | 1541 ±10 |
| 4 | Opus 4.7 | high | 1540 ±5 |
| 5 | Fable 5 | high | 1538 ±6 |
| 6 | Opus 4.6 | high | 1537 ±5 |
| 7 | Opus 4.6 | max | 1534 ±5 |
| 8 | Opus 5.5 | high | 1533 ±9 |
| 9 | Muse Spark 1.2 | xhigh | 1532 ±9 |
| 10 | Opus 4.7 | max | 1531 ±5 |
| 11 | GPT-6.1 Sol | max | 1529 ±10 |
| 12 | MiMo V2.6 Pro | max | 1528 ±15 |
| 13 | Muse Spark 1.3 | max | 1526 ±7 |
| 14 | GPT-6 Astra | max | 1526 ±8 |
| 15 | Muse Spark 1.1 | max | 1525 ±6 |
| 16 | Fable 5.1 | max | 1525 ±8 |
| 17 | Opus 4.8 | high | 1523 ±5 |
| 18 | Muse Spark | 1523 ±8 | |
| 19 | Opus 5 | high | 1522 ±5 |
| 20 | Kimi K3 | max | 1522 ±6 |
| 21 | GPT-5.6 Sol | xhigh | 1521 ±6 |
| 22 | Gemini 3.8 Flash | high | 1521 ±6 |
| 23 | Kimi K3.1 | max | 1519 ±17 |
| 24 | Opus 5 | max | 1517 ±6 |
| 25 | Opus 4.8 | max | 1516 ±5 |
| 26 | GLM 5.3 | max | 1515 ±7 |
| 27 | GLM 5.3 Flash | max | 1515 ±6 |
| 28 | DeepSeek V4.1 Flash | max | 1514 ±9 |
| 29 | Sonnet 4.6 | max | 1514 ±5 |
| 30 | Qwen 3.8 Max | max | 1514 ±6 |