The web, as clean Markdown.

One request turns a page, a PDF or a document into model ready text. Deterministic, with no model in the loop, on the same key as everything else.

reader

You can point the reader at:

Everything it returns is the same clean Markdown, on the same key, billed from the same balance.

See the rates

Any page on the web

Fetch a URL and get the article back as Markdown. Navigation, cookie banners, ad slots and scripts are dropped; headings, links and code survive as themselves.

PDFs and documents

PDF and DOCX go through the same converter as HTML, so a report becomes the same structured text every time, tables included.

A whole site, indexed

Point the indexer at a docs site and every page it reads becomes embedded, searchable context at €0.20 per million tokens.

Search, then read

One call returns ranked results, another fetches the pages behind them. Deep research runs both in a loop and hands back what it found.

request
curl https://api.nemu.cc/reader \
-G --data-urlencode "url=https://example.com/post" \
-H "Authorization: Bearer $NEMU_KEY"
200 · text/markdown · 1 page read
# The post title
Clean paragraphs, headings and links:
no nav, no cookie banners, no scripts.
## A section survives as a section

Deterministic by design

No model in the loop. HTML, PDF and DOCX are fetched and converted to clean Markdown the same way every time, which keeps the base price low.

Billed like everything else

Reader calls draw from the same credit balance as the rest of the platform. No separate subscription, no separate key.

Bring your own model on top

Want a summary of what you read? Pipe the Markdown to any mapped model, and that call runs on your keys, free of credits.

EndpointWhat it doesCost
GET /reader?url=…Fetch a page and return clean Markdown1 page read · €0.003
POST /reader/searchRun a web search, return ranked results1 web search · €0.006
POST /reader/deep-searchMulti-page research with content fetched1 deep research run · ~€0.05