MobiFlow: 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. Success rate in the base scenario, the share of tasks whose goal state is reached, for general multimodal models in the shared framework of the paper, which marks actionable screen elements with OmniParser, in percent; higher is better.
Metric: Success rate (%) in the base scenario, the share of tasks whose goal state is reached 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; general models in the paper's shared execution framework (OmniParser marks actionable elements; OpenRouter, temperature 0); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 49.7 | #178 |
| 2 | Claude Opus 4.1 | 43.6 | #278 |
| 3 | Gemini 3 Flash | 40.2 | #175 |
| 4 | Gemini 2.5 Pro | 38.7 | #276 |
| 5 | Gemini 2.5 Flash | 30.7 | #463 |
| 6 | Grok 4 | 8.3 | #319 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=mobiflow · How It Works · Data refreshed daily, snapshot 2026-11-21.