Updated daily at 05:00 UTC
Benchmark
ScreenSpot-Pro
Measures: Given a high-resolution screenshot of professional software (CAD, creative, scientific, dev tools, office) and an instruction, the model must point to the target UI element; scored as % of clicks landing inside the correct element's box.
- Maintainer
- Kaixin Li, Ziyang Meng et al., National University of Singapore and collaborators
- Introduced
- 2025
- Status
- Active
- Saturation
- Saturated
- Typical provenance
- Self-reported
- In scoring set
- No
Leaderboard
Independent and third-party results only.
No independent or third-party results yet. Show all results