New

• Added option to use tools by default
• Redesign model manager
• Added option to attach audio and video files (audio transcription)
• Added Spotify controller tool

Fixes

• Fixed thinking mode being on bugs out normal models

Fixes

• Fixed Alpaca crashing when chats database has incorrect metadata (type)
• Fixed Quick Ask and Live Chat not running when opened from terminal whilst Alpaca is running
• Fixed stop button not working when pressed before text generation beggins
• Fixed instances sometimes not appearing

New

• Live Chat (Talk to models as if you were on a call)
• Live message rendering
• Faster dictation
• Dictation whilst message is being generated
• Camera picture attachment
• Quick Ask now supports attachments
• Separator element for message and attachment rendering
• Redesigned popups
• Thought switch for Ollama instances
• Username sharing for Ollama instances
• Redesign for instance manager and tool manager
• Better web searching tool with options

Fixes

• Fixed duplication of thought attachments
• Fixed weird behavior with model selector
• Ollama (Managed) instances behave better
• Better optimizations for RAM (~2GB difference)
• Instant chat rendering
• Models profile pictures are now saved at a higher resolution
• Quick Ask messages can now be sent by using the enter key
• Alpaca is less prone to crashes now

Fixes

• Fixed new chat not being created when the chat list is empty

Fixes

• Chat exporting now works as expected
• Fixed automatic creation of Ollama instance
• Added Hebrew and Telugu credits
• Fixed STT model selector

New

• Multiple QuickAsk window rendering

Fixes

• Better stability for QuickAsk
• Creating a new chat now selects it

New

• Chat Search
• Hide "latest" and "custom" tags from model selector
• Hide model's languages behind popup
• New models listed for Ollama
• Added option to autodictate new mesages
• Added Meta Llama API to list of instances
• Added new env variables options ("24H hour formatting" and "only Ollama mode")
• Made Mermaid scripts executable using Python HTTP server

Fixes

• Better stability when switching instances
• Better performance when navigating menus
• Fixed some dialogs not appearing in Quick Ask window
• Faster message search
• Faster message rendering

Fixes

• Fixed whisper directory not existing causing error

New

• Updated model list
• Alpaca now remembers it's size
• Added reasoning category for Ollama models

Fixes

• Improvements in sample prompts

Fixes

• Fixed auto creation of Ollama (Managed) instance
• Removed legacy JSON to SQLite3 migration code
• Fixed power saving mode appearing whilst using online instances

Fixes

• Fixed Ollama (Manged) instance not being able to be created

Fixes

• Instance manager now follows default model
• English text-to-speech voices not working
• Instance manager sometimes not saving instances
• Fixed Gorq and Deepseek instances not generating text

New

• Smart tools for models
• Speech recognition (message dictation)
• Text to Speech
• New Quick Chat system
• Filter Ollama models by categories
• Better math Latex rendering in messages
• Rich text rendering in attachment preview
• Matplotlib is now included in Python code runner
• Styling for messages being generated

New Instances

• Deepseek
• OpenRouter AI
• Anthropic
• Groq Cloud
• Fireworks AI
• Lambda Labs

New Attachment Types

• Microsoft Word Document (docx)
• Microsoft PowerPoint Document (pptx)
• Microsoft Excel Document (xlsx)

New Tools

• Run Command (Testing)
• Online Search
• Extract Wikipedia Article
• Get Recipe by Name
• Get Recipes by Category
• Get Current Datetime

Fixes

• Fixed welcome screen not showing sometimes when deleting a message
• Fixed bold text not rendering correctly in tables
• Fixed sample prompt buttons overflowing on small screens

New

• Updated runtime to Gnome 48
• Added back 'category pills' to model manager
• Better appearance for model manager sidebar
• New models

Fixes

• Fixed bad title generation with chain-of-thought models
• Hide page switcher in model manager if there's only one page (online instances)

New

• Option to delete all chats
• Button to refresh sample prompts
• New models

Fixes

• Fixed saving Ollama (Managed) instance might crash it
• Fixed stop button
• Fixed model search not working if there are only pulling models
• Fixed sample prompts sometimes not appearing

Fixes

• Don't clear the building output of C++ scripts
• Better handling of attachments
• Handle remote Ollama instance's API Key better
• Remove '\n' characters in instance edit page

New

• Dynamic chat loading
• Updated Ollama instance

New

• Tweaked appearance of models in model manager
• Updated Ollama instance to 0.5.11
• Added new models

New

• New instance manager
• New welcome screen

New Instances

• OpenAI ChatGPT
• Google Gemini
• Together AI
• Venice

Fixes

• Exporting chats with 'thoughts' attachment is fixed
• Fixed attachment filters

New

• New model manager
• Changed GtkSpinner to AdwSpinner
• Better handling of launch process
• New loading screen at launch
• Better handling of file types
• Better regex expression for LaTeX equations
• Confirmation dialog if user closes Alpaca whilst a model is downloading
• Better handling of think tags in messages
• Default model is now in charge of generating titles
• Message header is now shown whilst the message is being generated
• Better handling of model profile pictures
• New models in 'available models' list

New

• Added option for attaching screenshots
• Basic LaTeX math equations are now rendered in messages
• HTML and C++ scripts can now be run inside Alpaca
• Added option to open the environment directory from the terminal
• Added option to edit code blocks directly
• Complete keyboard shortcut list
• Images are now attached in 640p resolution
• Website attachments now use extracted titles
• Better chat title generation
• Added option to attach any plain text files
• Added spellchecker to message entry
• Alpaca's parameters are now stored using SQLite3
• Small appearance changes in text entries

Fixes

• Alpaca's launch process is more reliable
• Closing the terminal now kills the script subprocess

New

• Transitioned chat backend from JSON to SQLite3
• Changed appearance of messages
• Added the option to add profile pictures to models

Fix

• Changed override HIP_VISIBLE_DEVICES to ROCR_VISIBLE_DEVICES

New

• Added categories to models
• Specified model's languages
• Added warning when downloading embedding models

Fixes

• Replaced low ram warning with big model warning

Fixes

• Correctly escape markup before rendering message
• Fixed about dialog not working if log file was missing

New

• System messages can now be sent directly from Alpaca
• New redesign for messages and smaller minimum size
• New models included in 'available models list'
• Added symbolic icon when attaching code files
• When exporting a chat it now includes a markdown file
• Refresh button in model manager when using a remote instance
• Assistant messages are now editable
• Updated Ollama to v0.5.2
• New option to change model directory
• File previewer now resizes dynamically to content
• Adapted Alpaca to work without integrated Ollama instance
• Compatibility added with ODT files

• Restored ROCm compatibility
• Added long press gesture to chat rows so actions can be done in touchscreens
• Fixed edit button not saving changes
• Changed max temperature value to 2
• Made seed 0 actually random
• Fixed Gnome search provider not working outside of Flatpak installations

New

• New option --ask MESSAGE, to open a new 'Quick Ask' window
• Gnome Search integration now works whilst the app is opened

New

• Added launch paramenters --ask MESSAGE, --new-chat CHAT, --select-chat CHAT, --list-chats, --version
• Added integration as Gnome Search Provider
• Updated Ollama to v0.4.2 with new models

New

• User messages are now compacted into bubbles

Fixes

• Fixed re connection dialog not working when 'use local instance' is selected
• Fixed model manager not adapting to large system fonts

New

• Details page for models
• Model selector gets replaced with 'manage models' button when there are no models downloaded
• Added warning when model is too big for the device
• Added AMD GPU indicator in preferences

New

• Better system for handling dialogs
• Better system for handling instance switching
• Remote connection dialog

Fixes

• Fixed: Models get duplicated when switching remote and local instance
• Better internal instance manager

New

• Added 'Cancel' and 'Save' buttons when editing a message

Fixes

• Better handling of image recognition
• Remove unused files when canceling a model download
• Better message blocks rendering

New

• Run bash and python scripts straight from chat
• Updated Ollama to 0.3.12
• New models!

Fixes

• Fixed and made faster the launch sequence
• Better detection of code blocks in messages
• Fixed app not loading in certain setups with Nvidia GPUs

Fixes

• Fixed message notification sometimes crashing text rendering because of them running on different threads

Fixes

• Fixed message generation sometimes failing

New

• Sidebar resizes with the window
• New welcome dialog
• Message search
• Updated Ollama to v0.3.11
• A lot of new models provided by Ollama repository

Fixes

• Fixed text inside model manager when the accessibility option 'large text' is on
• Fixed image recognition on unsupported models

Fixes

• Fixed spinner not hiding if the back end fails
• Fixed image recognition with local images
• Changed appearance of delete / stop model buttons
• Fixed stop button crashing the app

New

• Made sidebar resize a little when the window is smaller
• Instant launch

Fixes

• Fixed error on first run (welcome dialog)
• Fixed checker for Ollama instance (used on system packages)

Fixes

• Fixed 'clear chat' option
• Fixed welcome dialog causing the local instance to not launch
• Fixed support for AMD GPUs

New

• Model, message and chat systems have been rewritten
• New models are available
• Ollama updated to v0.3.9
• Added support for multiple chat generations simultaneously
• Added experimental AMD GPU support
• Added message loading spinner and new message indicator to chat tab
• Added animations
• Changed model manager / model selector appearance
• Changed message appearance
• Added markdown and code blocks to user messages
• Added loading dialog at launch so the app opens faster
• Added warning when device is on 'battery saver' mode
• Added inactivity timer to integrated instance

• The chat is now scrolled to the bottom when it's changed
• Better handling of focus on messages
• Better general performance on the app

New

• New duplicate chat option
• Changed model selector appearance
• Message entry is focused on launch and chat change
• Message is focused when it's being edited
• Added loading spinner when regenerating a message
• Added Ollama debugging to 'About Alpaca' dialog
• Changed YouTube transcription dialog appearance and behavior

Fixes

• CTRL+W and CTRL+Q stops local instance before closing the app
• Changed appearance of 'Open Model Manager' button on welcome screen
• Fixed message generation not working consistently
• Fixed message edition not working consistently

New

• Model manager opens faster
• Delete chat option in secondary menu
• New model selector popup
• Standard shortcuts
• Model manager is navigable with keyboard
• Changed sidebar collapsing behavior
• Focus indicators on messages
• Welcome screen
• Give message entry focus at launch
• Generally better code

Fixes

• Better width for dialogs
• Better compatibility with screen readers
• Fixed message regenerator
• Removed 'Featured models' from welcome dialog
• Added default buttons to dialogs
• Fixed import / export of chats
• Changed Python2 title to Python on code blocks
• Prevent regeneration of title when the user changed it to a custom title
• Show date on stopped messages
• Fix clear chat error

New

• Changed shortcuts to standards
• Moved 'Manage Models' button to primary menu
• Stable support for GGUF model files
• General optimizations

Fixes

• Better handling of enter key (important for Japanese input)
• Removed sponsor dialog
• Added sponsor link in about dialog
• Changed window and elements dimensions
• Selected model changes when entering model manager
• Better image tooltips
• GGUF Support

New

• Regenerate any response, even if they are incomplete
• Support for pulling models by name:tag
• Stable support for GGUF model files
• Restored sidebar toggle button

Fixes

• Reverted back to standard styles
• Fixed generated titles having "'S" for some reason
• Changed min width for model dropdown
• Changed message entry shadow
• The last model used is now restored when the user changes chat
• Better check for message finishing

New

• Added table rendering (Thanks Nokse)

Fixes

• Made support dialog more common
• Dialog title on tag chooser when downloading models didn't display properly
• Prevent chat generation from generating a title with multiple lines

New

• Bearer Token entry on connection error dialog
• Small appearance changes
• Compatibility with code blocks without explicit language
• Rare, optional and dismissible support dialog

Fixes

• Date format for Simplified Chinese translation
• Bug with unsupported localizations
• Min height being too large to be used on mobile
• Remote connection checker bug

Fixes

• Models with capital letters on their tag don't work
• Ollama fails to launch on some systems
• YouTube transcripts are not being saved in the right TMP directory

New

• Debug messages are now shown on the 'About Alpaca' dialog
• Updated Ollama to v0.3.0 (new models)

Fixes

• Models with '-' in their names didn't work properly, this is now fixed
• Better connection check for Ollama

Stable Release

The new icon was made by Tobias Bernard over the Gnome Gitlab, thanks for the great icon!

Features and fixes

• Updated Ollama instance to 0.2.8
• Better model selector
• Model manager redesign
• Better tag selector when pulling a model
• Model search
• Added support for bearer tokens on remote instances
• Preferences dialog redesign
• Added context menus to interact with a chat
• Redesigned primary and secondary menus
• YouTube integration: Paste the URL of a video with a transcript and it will be added to the prompt
• Website integration (Experimental): Extract the text from the body of a website by adding it's URL to the prompt
• Chat title generation
• Auto resizing of message entry
• Chat notifications
• Added indicator when an image is missing
• Auto rearrange the order of chats when a message is received
• Redesigned file preview dialog
• Credited new contributors
• Better stability and optimization
• Edit messages to change the context of a conversation
• Added disclaimers when pulling models
• Preview files before sending a message
• Better format for date and time on messages
• Error and debug logging on terminal
• Auto-hiding sidebar button
• Various UI tweaks

New Models

• Gemma2
• GLM4
• Codegeex4
• InternLM2
• Llama3-groq-tool-use
• Mathstral
• Mistral-nemo
• Firefunction-v2
• Nuextract

Translations

These are all the available translations on 1.0.0, thanks to all the contributors!

• Russian: Alex K
• Spanish: Jeffser
• Brazilian Portuguese: Daimar Stein
• French: Louis Chauvet-Villaret
• Norwegian: CounterFlow64
• Bengali: Aritra Saha
• Simplified Chinese: Yuehao Sui

Fix

Removed DOCX compatibility temporally due to error with python-lxml dependency

Big Update

• Added compatibility for PDF
• Added compatibility for DOCX
• Merged 'file attachment' menu into one button

Quick Fix

There were some errors when transitioning from the old version of chats to the new version. I apologize if this caused any corruption in your chat history. This should be the only time such a transition is needed.

Huge Update

• Added: Support for plain text files
• Added: New backend system for storing messages
• Added: Support for changing Ollama's overrides
• General Optimization

Big Update

• Added: Support for GGUF models (experimental)
• Added: Support for customization and creation of models
• Fixed: Icons don't appear on non Gnome systems
• Update Ollama to v0.1.39

Fix

• Fixed: app didn't open if models tweaks wasn't present in the config files

Big Update

• Changed multiple icons (paper airplane for the send button)
• Combined export / import chat buttons into a menu
• Added 'model tweaks' (temperature, seed, keep_alive)
• Fixed send / stop button
• Fixed app not checking if remote connection works when starting

Daily Update

• Added text ellipsis to chat name so it doesn't change the button width
• New shortcut for creating a chat (CTRL+N)
• New message entry design
• Fixed: Can't rename the same chat multiple times

The fix

• Fixed: Ollama instance keeps running on the background even when it is disabled
• Fixed: Can't pull models on the integrated instance

Quick tweaks

• Added progress bar to models that are being pulled
• Added size to tags when pulling a model
• General optimizations on the background

Quick fixes

• Fixed: Scroll when message is received
• Fixed: Content doesn't change when creating a new chat
• Added 'Featured Models' page on welcome dialog

Nice Update

• UI tweaks (Thanks Nokse22)
• General optimizations
• Metadata fixes

Quick fix

• Updated Spanish translation
• Added compatibility for PNG

New Update

• Updated model list
• Added image recognition to more models
• Added Brazilian Portuguese translation (Thanks Daimaar Stein)
• Refined the general UI (Thanks Nokse22)
• Added 'delete message' feature
• Added metadata so that software distributors know that the app is compatible with mobile
• Changed 'send' shortcut to just the return/enter key (to add a new line use shift+return)

Bug Fixes

• Fixed: Minor spelling mistake
• Added 'mobile' as a supported form factor
• Fixed: 'Connection Error' dialog not working properly
• Fixed: App might freeze randomly on startup
• Changed 'chats' label on sidebar for 'Alpaca'

Cool Update

• Better design for chat window
• Better design for chat sidebar
• Fixed remote connections
• Fixed Ollama restarting in loop
• Other cool backend stuff

Huge Update

• Added Ollama as part of Alpaca, Ollama will run in a sandbox
• Added option to connect to remote instances (how it worked before)
• Added option to import and export chats
• Added option to run Alpaca with Ollama in the background
• Added preferences dialog
• Changed the welcome dialog

Please report any errors to the issues page, thank you.

Yet Another Daily Update

• Added better UI for 'Manage Models' dialog
• Added better UI for the chat sidebar
• Replaced model description with a button to open Ollama's website for the model
• Added myself to the credits as the spanish translator
• Using XDG properly to get config folder
• Update for translations

Please report any errors to the issues page, thank you.

Quick Fix

• The last update had some mistakes in the description of the update

Please report any errors to the issues page, thank you.

Another Daily Update

• Added full Spanish translation
• Added support for background pulling of multiple models
• Added interrupt button
• Added basic shortcuts
• Better translation support
• User can now leave chat name empty when creating a new one, it will add a placeholder name
• Better scalling for different window sizes
• Fixed: Can't close app if first time setup fails

Please report any errors to the issues page, thank you.

Really Big Update

• Added multiple chats support!
• Added Pango Markup support (bold, list, title, subtitle, monospace)
• Added autoscroll if the user is at the bottom of the chat
• Added support for multiple tags on a single model
• Added better model management dialog
• Added loading spinner when sending message
• Added notifications if app is not active and a model pull finishes
• Added new symbolic icon
• Added frame to message textview widget
• Fixed "code blocks shouldn't be editable"

Please report any errors to the issues page, thank you.

Big Update

• Added code highlighting
• Added image recognition (llava model)
• Added multiline prompt
• Fixed some small bugs
• General optimization

Please report any errors to the issues page, thank you.

Fixes and features

• Russian translation (thanks github/alexkdeveloper)
• Fixed: Cannot close app on first setup
• Fixed: Brand colors for Flathub
• Fixed: App description
• Fixed: Only show 'save changes dialog' when you actually change the url

Please report any errors to the issues page, thank you.

0.2.2 Bug fixes

• Toast messages appearing behind dialogs
• Local model list not updating when changing servers
• Closing the setup dialog closes the whole app

Please report any errors to the issues page, thank you.

0.2.1 Data saving fix

The app didn't save the config files and chat history to the right directory, this is now fixed

Please report any errors to the issues page, thank you.

0.2.0

Big Update

New Features

• Restore chat after closing the app
• A button to clear the chat
• Fixed multiple bugs involving how messages are shown
• Added welcome dialog
• More stability

Please report any errors to the issues page, thank you.

0.1.2 Quick fixes

This release fixes some metadata needed to have a proper Flatpak application

0.1.1 Stable Release

This is the first public version of Alpaca

Features

• Talk to multiple models in the same conversation
• Pull and delete models from the app

Please report any errors to the issues page, thank you.