LMArena Vision Leaderboard: leaderboard
Extension of Chatbot Arena focused on multimodal models. Rankings for vision-language tasks based on human preference voting.
Metric: Arena Score. Source: arena.ai. Status: saturation imminent. 148 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 | 1313 |
| 2 | Claude Opus 4.7 (High) | 1301 |
| 3 | Qwen 3.8 Max | 1300 |
| 4 | Claude Opus 4.7 | 1299 |
| 5 | Claude Opus 4.6 (High) | 1299 |
| 6 | Muse Spark | 1294 |
| 7 | Claude Opus 4.6 | 1293 |
| 8 | Muse Spark 1.2 (xHigh) | 1292 |
| 9 | Claude Opus 5 (High) | 1290 |
| 10 | Gemini 3 Pro | 1289 |
| 11 | GPT-5.5 | 1286 |
| 12 | Gemini 3.6 Flash (High) | 1285 |
| 13 | Claude Opus 4.8 (High) | 1285 |
| 14 | Gemini 3.5 Flash (High) | 1284 |
| 15 | GPT-5.5 (High) | 1284 |
Interactive version: theaggregate.ai/benchmark?slug=lmarena-vision-leaderboard · How It Works · Data refreshed daily, snapshot 2026-09-05.