Skip to main content

​
Attaching Files

Use Add photos & files in the + menu, press C then U, drag files onto the composer, or paste them. Attachments ride with that prompt and are never stored on Nito’s side.
  • Images (PNG, JPEG, WebP) go to the model as images, so the model has to support vision. The browser resizes and strips metadata first: up to 10 images per turn, each up to 20 MiB, fitted within a 2048 px long edge.
  • Documents (text, Markdown, code, PDF, DOCX, PPTX, XLSX, XLS) are converted to plain text before sending, so they work on any model.
If the selected model cannot read an attachment, the composer says so and offers to open the picker filtered to compatible models.

​
Parser Mode

Each document chip shows a parser mode. The default is in Settings under Attachment parsing. On an Encrypted chat, Auto is hidden and Client is the default. Choosing Server shows the warning Server sees this file before encryption. Raw image uploads are not sent on Encrypted routes yet.

​
Free Plan

The free models today are text models, so attachments follow their capabilities. See Free Usage.

​
Next

Web Search and Tools

The live web.

Workbench

Save documents as Knowledge.

Privacy in the App

Parsing at Encrypted.

Choosing a Model

Vision and file support.