●Change8

v0.9.1

📦 ollama
✨ 7 features🐛 3 fixes🔧 6 symbols

Summary

This release introduces tool calling for DeepSeek-R1 and Magistral, alongside a major preview of native macOS and Windows applications featuring network exposure and custom model directories.

Migration Steps

  1. Run 'ollama pull' to update models for improved tool calling support.
  2. Download the new preview apps for macOS or Windows if testing the new application architecture.
  3. If disabling thinking mode on Magistral, update the system prompt as recommended.

✨ New Features

  • Tool calling support added for DeepSeek-R1-2508 and Magistral.
  • New preview versions of Ollama for macOS and Windows applications.
  • Network exposure: Ollama can now be exposed on the network for remote access.
  • Local browser access: Allows websites to access local Ollama installations via JavaScript libraries.
  • Custom model directory: Users can now modify the storage path for models (e.g., external disks).
  • Native macOS application: Smaller installation footprint and faster startup times.
  • Magistral model now supports disabling thinking mode.

🐛 Bug Fixes

  • Fixed issue on Windows where 'ollama run' would not start Ollama automatically.
  • Improved tool calling reliability for Llama 4 and Mistral.
  • Enhanced error messages to be more informative than the generic 'POST predict' error.

🔧 Affected Symbols

ollama runDeepSeek-R1-2508MagistralLlama 4Mistralollama-js