Benchmarks · coding
Arena WebDev · Image to WebDev
Human preference between agentic front-end web development outputs; Bradley-Terry rating. Only the Image to WebDev category.
Category
52 models · 57 results · data as of 2026-10-29 · arena.ai ↗
Leaderboard
Opus 5.5 max
All results
57 rows| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Opus 5.5 | max | 1753 ±24 |
| 2 | Sonnet 5.5 | xhigh | 1737 ±17 |
| 3 | GPT-6 Astra | max | 1726 ±12 |
| 4 | Fable 5.1 | max | 1722 ±13 |
| 5 | GPT-6.1 Sol | max | 1702 ±15 |
| 6 | Opus 5 | max | 1662 ±10 |
| 7 | GPT-6 Sol | max | 1660 ±12 |
| 8 | Muse Spark 1.3 | max | 1640 ±9 |
| 9 | Qwen 3.8 Max | max | 1637 ±18 |
| 10 | Fable 5 | high | 1623 ±9 |
| 11 | GPT-5.6 Sol | max | 1609 ±9 |
| 12 | DeepSeek V4.1 Flash | max | 1606 ±22 |
| 13 | Grok 4.6 | high | 1603 ±11 |
| 14 | Gemini 3.7 Flash | high | 1599 ±12 |
| 15 | MiMo V2.6 Pro | high | 1598 ±12 |
| 16 | Muse Spark 1.3 | xhigh | 1596 ±12 |
| 17 | MiMo V2.6 Flash | high | 1596 ±13 |
| 18 | Gemini 3.8 Flash | high | 1593 ±13 |
| 19 | Grok 4.7 | xhigh | 1586 ±17 |
| 20 | Kimi K3 | max | 1579 ±10 |
| 21 | GLM 5.3 Flash | max | 1578 ±9 |
| 22 | Opus 4.7 | high | 1577 ±9 |
| 23 | Qwen 3.8 27B | max | 1577 ±8 |
| 24 | GPT-6 Luna | max | 1571 ±10 |
| 25 | Opus 4.7 | max | 1562 ±9 |
| 26 | Grok 4.5 | max | 1562 ±10 |
| 27 | Gemini 3.6 Flash | high | 1545 ±14 |
| 28 | Opus 4.6 | high | 1543 ±9 |
| 29 | GPT-5.6 Terra | max | 1541 ±9 |
| 30 | Sonnet 5 | high | 1540 ±8 |