Skip to main content
Image input lets a model see. You attach a picture to your prompt and a model that supports vision reads it and answers. This is how you ask “what is in this screenshot,” “transcribe this receipt,” or “what does this chart show.” This page covers how to attach an image, the model requirement, and how documents differ from images.

​
Attaching an Image

Attach a picture with --file pointing at the image, and choose a model that supports images. --file is repeatable, so you can attach several pictures in one prompt.
You can run the same thing from your terminal with nito ask. See Ask and Fusion from the terminal. To attach more than one image, repeat --file:

​
Images Require an Vision-Capable Model

Not every model can read images. Images require a vision-capable model, and Nito checks this for you. If you attach an image to a model that cannot accept images, Nito refuses with a clear message rather than guessing, for example:
This is a feature: Nito will not silently send your image to the wrong model or drop it. To find a model that works, browse the catalog and pick one whose capabilities include vision:
Real vision-capable model IDs (for example an anthropic/claude-... vision model) appear in the catalog, but prefer describing what you need (“an vision-capable model”) and let the models list confirm the current options. See Choosing a model.

​
Images Versus Documents

An image is a picture that a vision model looks at. A PDF or office document is not an image: it is parsed to text first, so it works with any model, not just a vision one. If your file is a document rather than a picture, use the document path instead.

Multimodal Overview

What Nito accepts beyond text.

PDF and Document Input

Attach a PDF or office file and work with it using any model.

Choosing a Model

Browse the catalog and pick a model with the capabilities you need.