VLM2-Bench: leaderboard
Benchmark evaluating vision-language models on visually linking matched cues across multiple images and frames.
Metric: Average Score on VLM2-bench (9 subtasks). Source: huggingface.co. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 60.36 | #343 |
| 2 | Qwen 2.5 VL 7B | 54.82 | #643 |
| 3 | Qwen 2 VL 7B | 42.37 | #750 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vlm2-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.