v0.7.0

What's Changed

Context Injection

• Added automatic VSCode, Xcode, iTerm2 and terminal context injection

System Wide Dictation

• Launched on-device system-wide audio transcription using WhisperKit

Multimodality and Tool Use

• Added support for image inference, document upload
• Added link previews for web browsing

Offline Inference

• Added MLX backend support for on-device LLM generation

UI Improvements

• Refreshed overall interface design and animations
• Enhanced attachment and preview systems

v0.6.0

What's Changed

• Fixed bugs related to focused inputs (#6), model selection (#5) and other performance issues.
• Dropped the 'Mac'. Just HuggingChat. It's cleaner. (#8)
• Added basic local model inference support for GGUF with the ability to dynamically switch between local and server model using new keyboard shortcut ⌘+shift+\ (modifiable in Settings).
• Added a 'Components' pane in Settings for model download/import.
• Added the ability to import local text files (e.g. source code and text)
• Conversations can now be set to clear at preset intervals
• Added acknowledgments section to the About section highlighting OS tools used within the app.
• Fixed app update server.

What To Test

• Test local generation with various models, with a special focus on user experience.
• Please test model management (i.e. import, deletion, etc) in the dedicated Components pane.
• Import text/source code files to local and server-side models
• Test long context local inference

What's Next

• Codebase cleanup
• Add two more themes
• Quality of life improvements and stability fixes
• Open sourcing the code

HuggingChat v0.5.0

Initial beta release of the Hugging Chat MacOS application.

What To Test

• Login and onboarding flow: Make sure you are able to sign in to your account (i.e. w/ & w/o 2FA).
• Chat capabilities: Test model selection, responses, markdown parsing, etc.
• Customization: Test various customization options (e.g. code highlighting, general appearance, system theme)
• Performance: Interact with the models, note any lag or jitters, etc.

Known Issues

• Web browsing fails on the server-side
• Backend sometimes closes early without a response

Feedback

Please create an issue to submit app feedback!