✨ Alpaca 7: Live Chat, Cameras, Message Rendering and More! ✨

[Jeffser]

Hello everyone, I have the pleasure to present one of the biggest and most important Alpaca update!

Release Date: July 01

New

🎤 Live Chat

Talk to your models as if you were on a call with them.

Access this mode by going into the primary menu (where the instance and tool managers are)

📬 Live Message Rendering and Dictation

Messages are now rendered whilst they are being generated, every time the model finishes a line it gets fully rendered, that includes markdown, code, tables and all the other types of blocks.

Messages are now also dictated as they are being generated (It's an option in preferences)

🔊 Enhanced Dictation

In addition to live dictation, TTS is now much much faster, it's able to completely read a whole message without any weird pauses (given your system can keep up).

📷 Camera Picture Attachment

Want to ask questions about something physical you have? (Like your collection of small cacti)

Alpaca now supports taking pictures with your camera (or webcam)!.

📂 Quick Ask Supports Attachments

You can attach files by drag and drop or with the handy attachment button.

➖ Separator Message Block

Alpaca will now render separators (---) as actual separators, this is part of the global block rendering code so it applies to messages and attachments.

💻 Redesigned Instance and Tool Managers

The editors for instances and tools are now dialogs!

🤖 New Options for Ollama

You can now share your name with models and switch thought processing.

🔍 Better Web Search Tool

Now the search tool is actually useful!

It now uses DuckDuckGo's libraries to make full search requests, try not to ask the same thing over and over again, DDG might ban you for a couple of hours if you spam too much their services.

There are also now options for safe search and the number of max results.

✨ Redesigned Popups

A new and improved design for popups is here!

Fixes

• Fixed duplication of thought attachments
• Fixed weird behavior with model selector
• Managed Ollama instances now behave better
• Ultra optimizations for TTS, STT and message generation (~2GB) difference in best case.
• Instant chat rendering (no more animations tho)
• Model profile pictures are now saved at a higher resolution (Important for Live Chat)
• Alpaca is much much much more stable and less prone to crashes now (might still happen tho)

Backend

• There's now a checker that makes sure if Whisper (STT) and Kokoro (TTS) are available (not important for Flatpak)
• Preferences have been moved from SQLite to the GTK standard (GLib)
• Separated global footer (The thing at the bottom with the attachments and message box) to it's own widget.
• New system for YouTube attaching.

Final Message

Thank you all for your contributions, reports and suggestions, I'm happy to say that Alpaca's code base is much much better now, with all the testing I've done I've found that Alpaca is much more stable.

I hope I can focus on enhancing and making new features so that Alpaca can continue being one of the biggest local AI clients for Linux.

I'm also grateful to every single one of you that has donated to me. It also means the world to me that so many people enjoy using my app!

Also thank you for 150k downloads on Flathub!

I hope you guys enjoy the work I've put into this update.

[mags0ft]

You've done an awesome job, thanks for keeping on delivering!

[SpaciousCoder78]

Hi, @Jeffser . I have sent you an email about packaging alpaca to Debian. Can you please reply to it?