MobiFlow (GUI Models) - Action Match Rate: leaderboard

MobiFlow (2026) tests mobile GUI agents on 240 real-world tasks in 20 third-party Android apps, scoring progress on task graphs fused from real user trajectories. Action match rate in the instruction-following scenario, the share of step instructions the actions satisfy, for specialized GUI agent models in their own official frameworks, in percent; higher is better.

Metric: Action match rate (%) in the instruction-following scenario, the share of the task's step instructions the agent's actions satisfy on MobiFlow's 240 real-world tasks in 20 third-party Android apps, scored on task-completion state graphs fused from real trajectories, at most 50 steps; GUI agent models in their official execution frameworks (vLLM); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 4 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GUI-Owl-8B (MobiFlow checkpoint unspecified)65.4
2uitars-1.5-7B63.1#7957
3MobiMind-Mixed-4B59.9
4AutoGLM-Phone-9B30.5

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=mobiflow-gui-models-action-match-rate · How It Works · Data refreshed daily, snapshot 2026-11-22.