Benchmarks
Vision benchmarks
These benchmarks measure how well models understand images.
We track 265 benchmarks in this topic. Main benchmarks are listed by how many models they cover, most first. Where a benchmark has separate results for a subset of its tasks, those results appear right under it. Click one for its leaderboard.
Showing 101–200 of 265 benchmarks.
- DocVQAtest7 models
- Vibe-Eval7 models
- InfoVQA6 models
- MMBench6 models
- MMMU-Pro (with tools)5 models
- BLINK - Art_Style4 models
- BLINK - Counting4 models
- BLINK - Forensic_Detection4 models
- BLINK - IQ_Test4 models
- BLINK - Jigsaw4 models
- BLINK - Multi-view_Reasoning4 models
- BLINK - Object_Localization4 models
- BLINK - Relative_Depth4 models
- BLINK - Relative_Reflectance4 models
- BLINK - Semantic_Correspondence4 models
- BLINK - Spatial_Relation4 models
- BLINK - Visual_Correspondence4 models
- BLINK - Visual_Similarity4 models
- DUDE4 models
- LiveSports-3K4 models
- MMMU (validation)4 models
- OVBench4 models
- OVOBench4 models
- BrowserGym VisualWebArena3 models