AndroidWorld: leaderboard
AndroidWorld evaluates mobile GUI agents on interactive Android tasks that require browser, desktop, mobile, or app-style environment control.
Metric: Success Rate pass@1 (%). Source: github.com. Saturation forecast: Around January 2027. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Sol | 100 | #14 |
| 2 | Gemini 3.5 Flash | 93.1 | #60 |
| 3 | Median Human | 80 | |
| 4 | MAI-UI-8B | 70.7 | #474 |
| 5 | GUI-Owl-7B | 66.4 | #852 |
| 6 | GPT-4o | 47.4 | #343 |
| 7 | GPT-4 Turbo | 30.6 | #404 |
| 8 | Qwen 2.5 VL 7B | 27.6 | #643 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=androidworld · How It Works · Data refreshed daily, snapshot 2026-10-20.