Chatbot Arena (Vision): leaderboard
The user uploads an image together with a text prompt, two anonymous models answer, and the user votes for the better answer. Ratings use the same Bradley-Terry fit as the text arena applied to these image battles, taken from the site's default view, which controls for answer length and markdown formatting. Higher is better, and the board covers only models that accept image input.
Metric: Arena Score. Source: arena.ai. Saturation forecast: Around June 2027. 165 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5 (High) | 1308 | #16 (Claude Fable 5) |
| 2 | Qwen 3.8 Max | 1301 | #35 |
| 3 | Claude Opus 4.6 (High) | 1299 | #63 (Claude Opus 4.6) |
| 4 | Claude Opus 4.7 | 1298 | #49 |
| 5 | Claude Opus 4.7 (High) | 1298 | #49 (Claude Opus 4.7) |
| 6 | Gemini 3.7 Flash (High) | 1297 | #23 (Gemini 3.7 Flash) |
| 7 | Claude Opus 4.6 | 1294 | #63 |
| 8 | GPT-6.1 Sol (Max) | 1294 | #7 (GPT-6.1 Sol) |
| 9 | Muse Spark | 1294 | #64 |
| 10 | Muse Spark 1.2 (xHigh) | 1293 | #62 (Muse Spark 1.2) |
| 11 | Claude Opus 5.5 (High) | 1293 | #3 (Claude Opus 5.5) |
| 12 | Muse Spark 1.3 (Max) | 1290 | #24 (Muse Spark 1.3) |
| 13 | Gemini 3 Pro | 1289 | #80 |
| 14 | Gemini 3.8 Flash (High) | 1289 | #21 (Gemini 3.8 Flash) |
| 15 | GPT-5.6 Sol (xHigh) | 1287 | #19 (GPT-5.6 Sol) |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=chatbot-arena-vision · How It Works · Data refreshed daily, snapshot 2026-10-20.