MobileGym-Bench - Cross-App Transfer Tasks: leaderboard

Metric: Success rate (%; 56 of the 256 test tasks, state-checked; open models: mean of four runs with re-sampled task parameters; proprietary models: one run (two for Gemini 3.1 Pro)). Source: arxiv.org. Saturation forecast: Around May 2027. 9 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)33
2Seed 2.0 Pro25
3Qwen 3.6 Plus16.1
4uitars-1.5-7B2.7
5Qwen 3 VL 4B Instruct0.9

Interactive version: theaggregate.ai/benchmark?slug=mobilegym-bench-cross-app-transfer-tasks · How It Works · Data refreshed daily, snapshot 2026-09-25.