In OS 27, can the Apple Intelligence models within shortcuts take images, or only text?

In Shortcuts, Apple Intelligence models are not limited to text-only input.

From the newer 2025 Keynote and Develop for Shortcuts and Spotlight with App Intents, the Use Model action lets you tap into Apple Intelligence models in shortcuts for tasks like parsing text or formatting data, and it can also use ChatGPT if needed. However, the context there only explicitly describes text-based use cases.

The clearest answer about images comes from What’s new in the Foundation Models framework: the on-device model is gaining vision capabilities, and you can insert an image attachment into your prompt together with text so the model can answer questions about the image. It supports image attachments from types like UIImage, NSImage, CGImage, and file URLs.

So:

• Apple Intelligence models in Shortcuts can work with images, not just text — at least in the newer model APIs with vision support.
• The shortcut-specific context explicitly highlights text tasks, but the broader Foundation Models update confirms image input support.