Change8

v4.2.2

📦 localaiView on GitHub →
1 features🐛 2 fixes🔧 2 symbols

Summary

This release includes bug fixes for VRAM parsing and Ollama endpoint parity, alongside dependency updates and new spec-decoding options exposed through the llama-cpp backend.

✨ New Features

  • Exposed new spec-decoding options via llama-cpp integration (bump to commit 1ec7ba0c) and adapted grpc-server.

🐛 Bug Fixes

  • Fixed parsing of Vulkan VRAM from text.
  • Ollama integration now accepts the `prompt` alias on the /api/embed endpoint for parity.

Affected Symbols