AndroidDaily: leaderboard

AndroidDaily (2026) tests mobile GUI agents on 350 everyday tasks in 94 real closed-source Android apps, from food delivery and shopping to social media and travel. Each session is judged against task-specific obligations, quality criteria and forbidden actions; the score is the task success rate in percent; higher is better.

Metric: Task success rate (%) over all 350 tasks in 94 real closed-source Android apps, end-to-end sessions judged by the GRADE evaluator (Gemini 3 Pro backbone, 87.37 percent agreement with human review) against three-tier task guidelines; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash62#175
2Gemini 3 Pro58.6#145
3Seed 1.846.3#252
4Seed 2.0 Pro45.4#222
5UI-TARS-1.542.3
6GLM-4.6V33.1#710
7Step-GUI-30A329.7
8Step-GUI-8B29.1
9UI-Venus-1.5-8B22.9
10GUI-Owl-1.5-32B-Think21.6
11GUI-Owl-1.5-8B-Think19
12MAI-UI-8B9.7

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=androiddaily · How It Works · Data refreshed daily, snapshot 2026-11-21.