Lumen
Lumen has been updated, please send email to help@ncsa.illinois.edu with subject Lumen if you have questions. Certain models will need to be acknowledged before use (one time only).

Chat

The Chat page (/chat) is the main interface for interacting with AI models.

Chat page

Page Layout

  • Left sidebar — Lists your conversations. Click any conversation to reload its message history.
  • Model selector — At the top of the chat area. Choose which model to send your message to. Graylisted models show a warning (⚠) if you haven't acknowledged them yet.
  • Chat area — Displays message bubbles. Your messages appear on the right; assistant replies appear on the left.
  • Message detail — Each assistant message has an info icon (ⓘ). Click it to see the token counts, response duration, and output speed for that reply.

Conversation Management

  • New conversation — Click the + New button in the sidebar.
  • Switch conversations — Click any conversation in the sidebar to load its history.
  • Remove conversation — Hover over a conversation and click the ✕ button.
  • Conversation title — Automatically generated from your first message.

Sending Messages

  1. Select a model from the dropdown at the top.
  2. Type your message in the input area at the bottom.
  3. Press Enter to send. Hold Shift+Enter to insert a newline without sending.

Streaming and Thinking

Responses stream in character by character as the model generates them. For models that support reasoning, a collapsible Thinking… section appears above the final answer, showing the model's chain-of-thought.

File Attachments

Attach a file to your message in two ways:

  • File picker — Click the paperclip (📎) button.
  • Drag and drop — Drag a file directly onto the message input bar.

Supported file types:

Type Extensions
Documents .txt, .md, .csv, .json, .py, .js, .ts, .html, .css, .xml, .yaml, .yml
PDFs .pdf
Images .png, .jpg, .jpeg, .gif, .webp
  • Images are sent as part of a "see and answer" message, for models that can interpret images.
  • Documents are read and prepended to your message text.
  • PDFs are parsed and appended as text.

An attachment chip appears above the input while composing. Click ✕ on the chip to remove it before sending.

Markdown and Math

Assistant responses are rendered with full formatting support:

  • Markdown — Headers, code blocks, lists, links, tables, and more.
  • Math — Inline $…$ and display $$…$$ LaTeX expressions rendered with KaTeX.
  • Code blocks — Dollar signs inside fenced code blocks are preserved and not treated as math.

On desktop, click the ←/→ arrow button in the header to collapse or expand the conversation sidebar. On mobile, the sidebar opens as an overlay via the hamburger menu.

National Center for Supercomputing Applications

Lumen

Illinois Computes GitHub Repository Request Feature