vmount 1.3.2
Local AI models and more reliable file tasks.
Local AI model support
Use Ollama, LM Studio, or another OpenAI-compatible model server with the AI assistant. Configure the endpoint, model, and context size in AI Settings. An API key is optional when your server does not require one.
The model server runs on your Mac or at your configured endpoint. File tools continue to run inside vmount’s VM. A server running on your Mac must accept connections from the VM.
Agent improvements
- Independent file reads can run together. Writes and other stateful operations run in order.
- Tool results stay attached to the correct call, including repeated calls to the same tool that finish out of order.
- Long conversations handle context compaction more reliably, with bounded recovery attempts when a model’s context is full.
Fixed file paths in AI chat
The assistant now receives the correct VM paths for the current folder, selected file, and file mentions. This prevents confusion between a browser path such as /btrfs-physical-0/… and its actual location under /mnt/.
Compatibility
Requires macOS 15 (Sequoia) or newer on an Apple Silicon Mac. The app and installer now use the same minimum OS requirement.