SVLAT - Image — leaderboard
Metric: Accuracy (%). Source: github.com. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 90.86 |
| 2 | Claude Opus 4.6 | 81.71 |
| 3 | GPT-5.4 | 75.71 |
| 4 | Qwen 3.5 9B | 69.43 |
| 5 | InternVL3.5-8B | 60.86 |
Interactive version: aibenchmarks.dev/benchmark?slug=svlat-image · How the rankings work · Data refreshed daily, snapshot 2026-07-18.