Versions

This page lists all Chat AI releases, starting with the newest.
Each entry describes new features, improvements, and fixes, with short explanations for how to use them.

v0.9.0 — September 2025

New Features

  • Redesigned UI
    A fresh interface with collapsible left and right sidebars for maximum chatting space.
    Optimized for handling arbitrarily large conversations and attachments smoothly.

  • New Model Selector
    Now located in the top center. Displays many more models compared to before.
    Default model switched to Qwen 3 30B A3B Instruct 2507, chosen for speed, tool compatibility, and strong performance.

  • GWDG Tools Integration
    Enable GWDG tools in the settings panel to unlock new capabilities:

    • Web search with Google
    • Image generation and modification
    • Speech generation (e.g., text-to-speech)
    • Arcana/RAG with any model
    • Custom MCP Server: specify the URL of any MCP server to access its tools in addition to GWDG tools.
      Chat AI displays real-time updates as tools are used.
      ⚠️ Tools are a new feature and may not yet work with all models. Recommended with Qwen 3 30B A3B Instruct 2507.

  • Export Data
    From your profile → settings, you can now backup all your data and save it as a single JSON file.

Improvements

  • Better code sanitization to prevent cross-site styling.
  • Smoother auto-scrolling in long conversations.
  • New navigation menu to other AI services.
  • Unified attach button for consistency.
  • Numerous small UI/UX fixes and optimizations.

v0.8.1 — July 3, 2025

New Features

  • Memory
    Chat AI can now remember relevant details across all conversations.
    ⚠️ Memory is stored locally in the browser only (not on servers).
    This allows more natural ongoing conversations, but clearing browser storage resets it.
  • Model search in selection menu
    Quickly find models by typing their name.
  • Configurable global timeout
    Prevents endless-loop responses by setting a max time per response.
  • LaTeX rendering option
    In addition to Markdown and Plaintext, responses can now render LaTeX properly.
  • Default settings in config file
    Define startup defaults (e.g., temperature, theme) in configuration.

Fixes

  • Fixed rendering bug in last line of text before references.
  • Generated arcana links now open in a new conversation.

v0.8.0 — June 4, 2025

New Features

  • Selectable Personas in UI
    Load personas directly in the interface from chat-ai-personas.
    Personas let you quickly change the assistant’s role.
  • Info messages in imported conversations/personas are now supported.
  • Standalone UI mode
    Simplified setup compatible with SAIA API key.

Improvements

  • Better import/export of conversations.

v0.7.4 — May 19, 2025

Fixes

  • Fixed syncing issues when multiple tabs are open.
  • Updated URL parsing for arcanas.

Improvements

  • References can now be rendered as links.

v0.7.3 — April 24, 2025

New Features

  • Sidebar settings panel replaces old settings popup.
  • Attachments as boxes with thumbnails in prompts.
  • Expanded attachment support for more file types.
  • Version number displayed in footer.

Changes

  • Removed key field from arcana settings.

v0.7.2 — April 15, 2025

New Features

  • Improved attachments display with resend, edit, undo support.
  • Added support for more file types.
  • Updated references format to align with RAG requirements.
  • Choose response rendering mode (Markdown, LaTeX, Plaintext).
  • Download responses as PDF.

Fixes

  • LaTeX, code, and Markdown rendering bugs fixed.
  • UI scrollbar issue fixed.
  • Multiple window/tab stability improved.

v0.7.1 — Feb 26, 2025

New Features

  • Video input support.
  • PDF processing via docling.
  • Image & text attachment previews.
  • Edit button for responses.

Improvements

  • Updated logo file.
  • Improved markdown/LaTeX rendering.
  • Smoother response display.
  • Better support for multiple conversations.

v0.7.0 — Feb 26, 2025

New Features

  • Arcanas supported.
  • Multiple conversation support.
  • Profile window added.
  • Scale-to-zero models supported.
  • Model status popups.
  • Code copy button for responses.

Fixes

  • Retry button path fixed.
  • LaTeX and Markdown handling improved.
  • Minor UI fixes.

v0.6.3 — Feb 26, 2025

New Features

  • No token limit – unlimited tokens.
  • Image upload – via clipboard or drag-and-drop.
  • CSV upload – supported directly.
  • Temporary model execution – run inactive models briefly.
  • Model status indicators (active, loading, etc.).

v0.6.2 — Feb 26, 2025

  • Updated models API endpoint to /models.

v0.6.1 — Feb 26, 2025

New Features

  • Share model & settings via base64-encoded URL.
  • Import external settings (e.g., personas).

UI Updates

  • More visible scrollbar in model selection.
  • Fixed header on tablets.
  • Cleaner design in options section.