ResearchResearch paperReinforcement Learning · Agents & Tool Use · Safety & Alignment1 source · Oct 9, 2026

OpenMobile-2: Building Versatile Mobile Agents with Scalable Environments and App-Native Tools

We introduce OpenMobile-2, a near-frontier mobile agent with fully open training environments and recipes.

Key points

  • We make three key advances: (1) Diverse environments with simulated commercial apps: We build MobileGym++, featuring 35 realistic, functionally rich commercial-style apps with cross-app workflows, while preserving full controllability for reset and verification.
  • (2) Open training data at scale: Building on this foundation, we curate nearly 12K mobile interaction trajectories for supervised fine-tuning and the largest open collection of verifiable RL training data for mobile agents, comprising over 2K executable tasks with automatic rewards.
  • (3) Hybrid GUI and app-native tool use: We explore an experimental mobile-use setting where apps expose selected functionalities as app-native tools alongside their GUIs, allowing agents to interleave GUI actions and tool calls within a task.
  • Additionally, we introduce MobileGym++ Bench for complex, long-horizon commercial mobile scenarios, supporting both GUI-only and hybrid GUI-tool evaluation on a shared task suite.

Sources (1)

  • [1]OpenMobile-2: Building Versatile Mobile Agents with Scalable Environments and App-Native Tools
    Hugging Face Daily Papers · Oct 9, 12:00 AM
    We introduce OpenMobile-2, a near-frontier mobile agent with fully open training environments and recipes.
    We make three key advances: (1) Diverse environments with simulated commercial apps: We build MobileGym++, featuring 35 realistic, functionally rich commercial-style apps with cross-app workflows, while preserving full controllability for reset and verification.

Extractive summary: sentences quoted from the sources.

Before this

  1. Oct 8, 2026openai/codex rust-v0.162.0: 0.162.0
  2. Oct 8, 2026Microsoft Joins the Local AI Push
  3. Oct 8, 2026Understanding Jev, the new model everyone is talking about
  4. Oct 8, 2026ConwayResearch/Underdog-Saluki-27B-1.0
  5. Oct 7, 2026[AINews] Claude Haiku 5.5 — better than GPT-6 Luna at the same pricing
  6. Oct 7, 2026Introducing Playground: Create and play custom games

Related