v4.8.1
📦 localaiView on GitHub →
✨ 3 features🐛 5 fixes🔧 3 symbols
Summary
This release includes several bug fixes for GGUF metadata, vLLM-cpp backend loading, and UI stability. It also adds new Qwen model variants to the gallery and enhances vLLM-cpp configuration.
✨ New Features
- Wire the full engine config surface through engine_args for vLLM-cpp.
- Add Qwen3.5 9B HauhauCS variants to the model gallery.
- Add Qwen3.5 9B Defiant Fable variants to the model gallery.
🐛 Bug Fixes
- Contain malformed GGUF metadata.
- Mirror the vLLM-cpp engine's ABI v10 so the backend loads again.
- Stop the traces page from crashing when switching trace tabs in the React UI.
- Restore 3D Studio results and history in the React UI.
- Ignore a half-populated socket activation environment in the CLI.