Releases · microsoft/Foundry-Local

Foundry Local Release Notes: v0.4 🚀

This release introduces several new features, bug fixes, and improvements to enhance your experience with Foundry Local.

New Chat Command: You can now start a new chat within an interactive session using the /new command. This will clear your current chat history, allowing for a fresh start.
Show License: We've added a foundry --license command to make it easy to display the license information for the Foundry CLI.

Force Download: A fix has been implemented for the foundry model download <model> --force command to ensure that any previous downloads are completely cleaned up.
Multi-GPU: A fix has been implemented where foundry model run <model>-cuda-gpu would error on a computer with more than one GPU card.

Delete All Models: You now have the ability to delete all models from the cache at once by specifying * as the model name.
Model Alias Support: We've expanded support for model aliases to the model unload and cache remove commands. This allows for the selection of the best model for your local machine, similar to how it works with model download, model load, and model run.
Remove All Models with Alias: The --all option has been added to the model unload and cache remove commands. This allows you to remove all models from the cache that match a specific model alias.
Interactive Run: The --interactive option has been removed from the run command. A run is now interactive by default unless a --prompt is specified.
Model ID Output: The model ID will now be displayed when downloading a model using an alias. This provides clarity on which model was selected for download.

The previous SDK packages are still the latest ones. Install them using: pip install foundry-local sdk or npm install foundry-local-sdk