The agent looks at images and PDF pages
9 November 2026
The agent can now look at the images and PDFs in your files, scanned pages included. Before, it read text files only and opened everything else through code.
- It looks at PNG, JPEG, GIF and WebP images, scaled down when large, and at up to 20 pages of a PDF at a time.
- Each look shows as a step of the answer with the file’s name and the pages read. A file that isn’t an image or a PDF is refused, and the step says why.