View only. Sign in to run this demo. You can also run it with your own model key, without an account.
Vision Extractor
Turn PDFs and images into validated, structured data in a single model call
How to use the Vision Extractor
- 1Add the documentsUpload PDFs and images, one or several. Or use one of the ready-made documents.
- 2Say what to extractDescribe the fields in words. The schema is generated for you to check.
- 3Choose the modelPick the model and options. Confidence scores per field are optional.
- 4Check the sourcesEach value shows the text it was read from, highlighted on its page.
- One model call sees every file at once, so it can match values across documents, for example an invoice against its purchase order.
- Nothing you enter is saved. A schema generated for your own task lives in this session only.
Start from a ready-made document
1. The documents
2. What to extract
A schema is generated from this text at the first run. Name the fields and their types.
Each value comes with its quote, highlighted in the PDF. Uses more output tokens.
No extraction yet
Every field of the result appears here, with the place in the document it was read from.
