Changelog

Connect securely to cloud providers with your own API key and control what you share.

• Local Only mode restricts inference to Ollama managed by Moose.
• Separate sharing permissions for messages and files. Review context before sending, allow messages per provider or per chat, and reset permissions in Privacy settings.
• Files in the sidebar brings attachments and library documents together. Browse, search, filter, preview, find the original folder, or reopen a related chat.
• More local models for chat, vision, reasoning, and coding.
• Clearer response status, with waiting and reasoning indicators that disappear as the answer arrives.
• HTTPS for external Ollama, with HTTP allowed only for connections on the same device.

Bring images and documents into your conversations, with local document search, source excerpts, and page references.

• Edit messages, regenerate answers, switch between saved versions, or continue in a separate chat.
• View model reasoning separately and use the options supported by your model.
• Keep drafts and attachments between sessions and move between chats during a response.
• Review download history, retry interrupted downloads, and chat while models download.
• Read mathematical formulas with local rendering and access to the original LaTeX.
• Use a wider chat layout, compact model controls, and clearer attachment previews.

• GPU acceleration through Vulkan for managed Ollama when available.
• A managed hardware preference to enable or disable GPU acceleration.
• The active acceleration backend appears in the provider status.

• Configurable shortcuts for chat, model, and navigation actions.
• A Shortcuts page in preferences to change, clear, or reset bindings.
• Conversation list indicators for the chat generating a response.

• Tools to organize conversation history and review older chats.
• Conversation export support to save chats outside Moose.

• Recent message history gives models conversation context.
• Per-chat system prompts, preferred models, context limits, and Ollama generation options.
• Built-in profiles and reusable custom profiles stored on your device.

• Markdown formatting for chat messages.
• Syntax highlighting, line numbers, and a copy action for code blocks.

The Flatpak can set up and manage Ollama without a host installation. Fresh installs can create an instance, download a model, and start chatting.

The first version of Moose introduced local Ollama chat, saved conversations, provider settings, and basic model management.