Starting a Conversation
To start a new chat:- Click New Chat in the sidebar or press Cmd/Ctrl + N
- Select a model from the model dropdown
- (Optional) Choose an assistant with pre-configured instructions
- Type your message in the input box
Sending Messages
Text Messages
Type your message in the chat input at the bottom of the screen:- Press Enter to send your message
- Press Shift + Enter to add a new line without sending
Attaching Images
For models with vision capabilities:- Click the + button in the chat input
- Select Add Images
- Choose one or more images (JPEG, JPG, PNG)
- Type your message and send
- Drag and drop images directly into the chat area
- Paste images from your clipboard (Cmd/Ctrl + V)
Vision features require models with vision capabilities. Jan will prompt you to download a vision model if needed.
Attaching Documents
For models with tool/RAG capabilities:- Click the + button in the chat input
- Select Add Documents
- Choose documents (PDF, DOCX, TXT, MD, CSV, XLSX, PPTX, HTML)
- Select ingestion mode:
- Inline - Insert document content directly into the prompt
- Embeddings - Create a vector database for semantic search
Message Actions
User Messages
For messages you send, you can:- Copy - Copy the message text to clipboard
- Edit - Modify your message and regenerate the response
- Delete - Remove the message from the thread
Assistant Messages
For AI responses, you can:- Copy - Copy the response to clipboard
- Regenerate - Generate a new response to the same prompt
- Edit - Modify the assistant’s message
- Delete - Remove the message from the thread
Advanced Features
Tool Calling
When a model supports tools, it can:- Search documents - Query embedded documents in the thread
- Execute MCP tools - Use Model Context Protocol servers
- Browser extension - Screenshot and analyze web pages
- Tool name and input parameters
- Execution status (running, success, error)
- Tool output or error message
Token Counter
Jan displays token usage and generation speed for each message:- Token count - Number of tokens in the current context
- Generation speed - Tokens per second during streaming
- Model capacity - Visual indicator of context usage
Reasoning Display
For models with reasoning capabilities (like DeepSeek R1), Jan shows the model’s thinking process:- Reasoning sections are collapsible
- Auto-expand during streaming
- Can be hidden for cleaner conversation view
Thread Management
Organizing Conversations
- Rename - Click the thread title to rename
- Delete - Remove individual threads or all threads
- Search - Press Cmd/Ctrl + K to search across threads
Projects
Organize threads into projects:- Create a new project from the sidebar
- Assign an assistant to the project
- Add project-specific documents
- All threads in the project share the same knowledge base
Project documents are accessible to all threads within that project using RAG tools.
Keyboard Shortcuts
| Action | Shortcut |
|---|---|
| New chat | Cmd/Ctrl + N |
| New project | Cmd/Ctrl + Shift + N |
| Toggle sidebar | Cmd/Ctrl + B |
| Search | Cmd/Ctrl + K |
| Send message | Enter |
| New line | Shift + Enter |
Stopping Generation
To stop an AI response while it’s generating:- Click the Stop button in the chat input area
- The partial response will be saved
- You can regenerate or continue the conversation
Context Management
If you receive a context length error:- Jan will display an Increase Context Size button
- Click it to automatically increase the model’s context window by 50%
- The model will restart with the new settings
- Your conversation will continue from where it stopped