v0.6.19
atlasia-ma/MoulSotv0.6.19Sep 10, 2026by RasulOs
Upgrading?
AI Summary
This release introduces bounded agent finalization and manual OAuth selection, alongside fixes for Android tap interactions and Langfuse span handling. It updates provider defaults to include the latest GPT-6, Gemini 3.8 Flash, and Claude Sonnet 5 models.
Key Highlights
- Bounded agent finalization
- Manual OAuth selection
- Updated provider defaults (GPT-6, Gemini 3.8, Claude Sonnet 5)
- Langfuse support for immutable ended spans
- Android tap fixes
New Features
- Bounded agent finalization
- Manual OAuth selection
- Langfuse support for immutable ended spans
- Updated provider defaults
Full Release Notes
Mobilerun 0.6.19 includes bounded agent finalization, manual OAuth selection and indexed Android tap fixes, refreshed provider defaults and compatibility, and Langfuse support for immutable ended spans. Provider defaults are GPT-6 Astra with low reasoning, Gemini 3.8 Flash (3.8 Flash Tiered for OAuth), and Claude Sonnet 5. This release pins Android Portal compatibility to 0.7.25. ## What's Changed * fix(agent): preserve task results during finalization by @lutz-grex in https://github.com/droidrun/mobilerun/pull/442 * fix: honor manual OAuth selection and avoid Android sibling tap obstructions by @RasulOs in https://github.com/droidrun/mobilerun/pull/443 * fix: refresh OAuth defaults and repair provider request compatibility by @RasulOs in https://github.com/droidrun/mobilerun/pull/444 * fix: normalize frozen Langfuse spans and update SDK test usage by @RasulOs in https://github.com/droidrun/mobilerun/pull/445 * chore: release mobilerun 0.6.19 by @RasulOs in https://github.com/droidrun/mobilerun/pull/446 **Full Changelog**: https://github.com/droidrun/mobilerun/compare/v0.6.18...v0.6.19
Related Repositories
rasbt/LLMs-from-scratch★ 107,081
Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
lyogavin/airllm★ 40,230
AirLLM 2.8T LLM inference with single 4GB GPU
NirDiamant/RAG_Techniques★ 29,962
This repository showcases various advanced techniques for Retrieval-Augmented Generation (RAG) systems. Each technique has a detailed notebook tutorial.
microsoft/OmniParser★ 25,539
A simple screen parsing tool towards pure vision based GUI agent
karpathy/nn-zero-to-hero★ 24,908
Neural Networks: Zero to Hero