Local path available · no account required

Ask your PDF a question. Keep the evidence attached.

N0X extracts and retrieves document text in your browser. With the Browser provider and network tools off, the document context and model prompt stay on this device.

Honest setup: local model weights are not bundled. Your first model choice downloads roughly 360 MB–2 GB for the recommended options, then initializes from cache on later visits.

Choose my document Try the sample first

retention-policy.pdf

Question

When is account data deleted?

Account data is deleted 30 days after closure. [retention-policy.pdf#chunk-3]

The citation identifies the exact file and retrieved chunk used as evidence.

Three steps, with the boundary visible

Choose your document

PDF, DOCX, Markdown, text, CSV, HTML, and JSON are processed in the browser.

Use a local provider

Browser models require a first-use weights download. The workbench shows that before it starts.

Verify the citation

Retrieved passages carry tags such as [policy.pdf#chunk-3]. If none qualifies, N0X supplies a no-evidence instruction.

When data can leave: Cloud API receives selected prompt and document context when you choose it. Deep Search sends the search query to configured search services. Image generation sends its prompt to the selected image route.