Image attachments, voice and read-aloud
Send images with a question, dictate a question, listen to an answer.
Images
The paperclip in the composer (or pasting / dropping an image into it) attaches images to the turn.
| Formats | PNG, JPEG, WebP, GIF |
| Limits | 6 images per turn, 10 MB each |
| Requirement | The selected model must be vision-capable. If not, the composer says so and you need to switch model. |
Images show as thumbnails in the question bubble. They are not indexed into the knowledge base — they only travel with that question. For anything you want to look up later, put the content in a document.
Voice input
The microphone button records; release and Ragenta transcribes it and appends to the composer (it does not send). Edit, then Enter. Needs speech-to-text on the deployment; otherwise the button is hidden.
Read aloud
The speaker button under an answer plays it as speech. Also only shown when the deployment has text-to-speech. Both count as the speech operation on the Usage page, limited to 30 per minute.