GMA - Atomic: leaderboard
Metric: Success rate (%; 300 tasks in seven custom Android applications, vision-only agent seeing raw screenshots (two most recent kept), MobileWorld action space plus call_user and answer, Qwen3.7-Plus user simulator, 150-step budget; 73 atomic single-action tasks). Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.8 Max | 93.15 |
| 2 | Seed 2.1 Pro | 89.04 |
| 3 | GPT-5.5 | 87.67 |
| 4 | Claude Opus 4.7 | 87.67 |
| 5 | GPT-5.6 Sol | 86.3 |
| 6 | Claude Sonnet 4.6 | 83.56 |
| 7 | Gemini 3.5 Flash | 80.82 |
| 8 | Qwen 3.7 Plus | 78.08 |
Interactive version: theaggregate.ai/benchmark?slug=gma-atomic · How It Works · Data refreshed daily, snapshot 2026-09-29.