- Text + vision with the active LLM; drop in images for multimodal prompts.
- Thinking mode — toggle the model's reasoning on/off per chat.
- Streaming responses with live token + reasoning display.
- Tabs — multiple conversations open at once; search your history.
- Per-message actions — copy, regenerate, edit-and-fork.
- Attachments — add files/images; chips stay clickable and are catalogued.
- Voice mode — exchange messages as voice notes (record + auto-play replies).
- Tools — built-in web tools (search, read URL, calculator, date/time); connector tools when Connectors is on.
- Scope — chat plainly ("Off Grid AI") or scoped to a project (grounded in its docs).