Repository navigation
AndroLLM v1.1.5 is here with major improvements to speed, stability, and the overall on-device AI experience. This release focuses on making the app feel smoother, faster, and more reliable across everyday use.
✨ What’s New
- 🎨 Refined the app experience with a more polished, premium feel
- ⚡ Faster and smoother local inference
- 🧠 Better model loading, switching, and runtime consistency
- 🔧 General app-wide improvements and cleanup
🚀 Performance Improvements
- Improved generation speed and responsiveness
- Better GPU/Vulkan execution behavior
- Reduced crashes and better memory handling
- More stable inference across repeated sessions
📦 Model Management
- Improved browsing for installed and available models
- Better backend/status reporting
- More reliable loading for supported local models
- Improved handling of compatibility and runtime issues
🛠 Fixes
- Fixed multiple UI inconsistencies
- Improved error handling for runtime and request failures
- Better retry behavior in unstable conditions
- More stable behavior during long AI sessions
🔍 Under the Hood
- Engine and runtime refinements
- Better diagnostics and logging
- Improved internal state handling
- General reliability improvements across the app
Thanks for using AndroLLM ❤️
More updates, optimizations, and features are on the way.