Skip to main content
A chat request can carry images for a vision-capable model and documents that Nito parses into text before the model runs. Both ride inside a message’s content array as typed parts: text, image_url, and file. One call can mix all three.
  • Image input: PNG, JPEG, WebP, as a URL, a data URL, or raw base64. The model must advertise image support.
  • Document input: PDF, DOCX, PPTX, XLSX, XLS, and plain text, parsed server-side.

Image Input

Images in chat.

PDF and Document Input

Documents in chat.

File Parser

How parsing works.