Chat

The chat view is the primary interface for interacting with local models in OMM. Messages stream in real-time with full markdown rendering, code syntax highlighting, LaTeX support, and multimodal input.

Sending Messages

Type your message in the input bar at the bottom and pressEnter to send. UseShift+Enter to add a new line without sending. The input bar expands automatically for longer messages and supports paste, drag-and-drop, and file attachments.

Multimodal Input

Attach images to your messages by clicking the paperclip icon or pasting from clipboard. OMM sends images directly to vision-capable models. Supported formats include PNG, JPEG, WebP, and GIF (first frame).

iNote
Vision support requires a multimodal model. Not all GGUF quantizations include vision capabilities. Check the model card on HuggingFace for details.

Streaming Responses

Responses stream token-by-token as the model generates them. A blinking cursor indicates active generation. You can stop generation mid-stream with the stop button or by pressing Esc.

Each streamed message shows the model name, token count, and generation time once complete. If the response is truncated due to context length limits, a continuation prompt appears.

Rich Content Rendering

Code Blocks

Fenced code blocks render with syntax highlighting for 100+ languages. Each code block includes a copy button in the top-right corner and a language label. Code blocks detect the language automatically when no fence label is specified.

LaTeX

Inline math wrapped in single dollar signs ($E = mc^2$) and display math in double dollar signs render via KaTeX. This works for any model that outputs LaTeX notation in its responses.

Tables

Markdown tables render with proper column alignment, alternating row colors, and horizontal scrolling for wide tables. Models that output comparison data in table format are rendered automatically.

Context Menu

Right-click any message to open the context menu with these actions:

  • Copy the full message text
  • Copy a specific code block
  • Edit your previous user message and regenerate the response
  • Regenerate the assistant response
  • Delete a message pair (user + assistant)
  • Branch the conversation from this point

Conversation Management

Branching

Edit a previous user message to branch the conversation. The original branch is preserved and you can switch between branches using the branch indicator that appears above the edited message.

System Prompt

Set a system prompt for the current conversation from the chat header. System prompts persist for the session and are included in the context sent to the model. You can save frequently used system prompts as part of a customization profile.

Scroll Management

The chat view auto-scrolls to the bottom during generation. If you scroll up to read earlier messages, auto-scroll pauses and a floating button appears to jump back to the latest content.

PreviousGGUF & SafeTensorsNextLibrary & Downloads