Apple trains 8B iOS UI navigation agents with online reinforcement learning
Overview
Apple Machine Learning Research says it trained UI navigation agents with online reinforcement learning directly on live iOS apps.
The framework splits the agent into a Planning agent and a UI Grounding model. According to the paper, the 8B RL agents reach nearly a sevenfold improvement over the supervised fine-tuning baseline when generalizing out-of-distribution to Android environments.
On real iOS applications, the authors report that the 8B agents outperform Gemini-3.5-Flash and tie Gemini-3.1-Pro. These results are the authors' own claims from their paper.
Written by AI from the articles below · updated Nov 2, 9:58 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Apple Machine Learning ResearchOfficialApple trains 8B UI navigation agents with online reinforcement learning on live iOS apps
AIApple Machine Learning Research says its online reinforcement learning framework trains UI navigation agents directly on live iOS apps, using a Planning agent paired with a UI Grounding model. The 8B RL agents reach nearly a sevenfold improvement over the SFT baseline in out-of-distribution generalization to Android, and outperform Gemini-3.5-Flash while tying Gemini-3.1-Pro on real iOS applications.
Heat Trend
- Current comparable heat
- 6
- Peak comparable heat
- 7Nov 2, 10:00 AM ET
- Change in past 24 hours
- –
The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.