ScreenSpot-Pro — leaderboard
Professional GUI grounding benchmark requiring agents to identify precise screen locations in high-resolution development, creative, and scientific software.
Metric: Grounding score (self-reported). Source: benchmarklist.com. Status: saturation imminent. 34 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Mythos Preview | 93 |
| 2 | Claude Mythos 5 | 90.7 |
| 3 | Claude Opus 4.8 | 89.5 |
| 4 | Claude Opus 4.7 | 87.6 |
| 5 | Qwen 3.7 Plus | 79 |
| 6 | Qwen 3.5 122B A10B | 70.4 |
| 7 | Qwen 3.5 27B | 70.3 |
| 8 | Qwen 3.5 35B A3B | 68.6 |
| 9 | Qwen 3.6 Plus | 68.2 |
| 10 | Gemini 3.1 Pro (Preview) | 68.1 |
| 11 | GPT-5.4 | 67.4 |
| 12 | Qwen 3.5 397B A17B | 65.6 |
| 13 | Qwen 3.5 9B | 65.2 |
| 14 | Qwen 3.5 4B | 60.3 |
| 15 | Qwen 3.5 2B | 54.5 |
Interactive version: theaggregate.ai/benchmark?slug=screenspot-pro · How the rankings work · Data refreshed daily, snapshot 2026-07-22.