v4.1.1
📦 localaiView on GitHub →
✨ 2 features🐛 6 fixes🔧 3 symbols
Summary
This patch release addresses regressions, primarily fixing Gemma 4 tokenization with llama.cpp and improving Anthropic API compatibility. It also includes UI updates and new features for coding agents.
✨ New Features
- Added thinking support for Gemma 4.
- Allow coding agents to interactively discover how to control and configure LocalAI.
🐛 Bug Fixes
- Fixed Gemma 4 tokenization when using llama.cpp.
- The UI now shows the login prompt when running in API key only mode.
- Improved error handling for NATS integration.
- Autoparser now prefers chat deltas emitted from backends.
- Fixed Anthropic API compatibility by showing null index when not present, defaulting to 0.
- Accumulated and stripped reasoning tags from autoparser results.