Summarise a PDF
Get the substance of a long document, using your own AI key.
Your PDF is never uploaded. The text is extracted in your browser, shown to you, and only sent after you approve it.
What this does
This extracts the document's text in your browser, sends it to the AI provider you choose using your own key, and writes the answer out as a Markdown file. The file itself never leaves your device, and the panel shows you the exact text and its size before anything is sent. There is no CekPDF server in that path, which also means no CekPDF account and no CekPDF bill - your provider charges you directly.
Use it when a document is longer than the time you have for it: a hundred-page tender you need to decide whether to bid on, minutes from a meeting you missed, an annual report where you only want the shape of the thing, or a contract you would like a map of before reading the clauses properly.
How it works
- Drop the PDF onto this page. Its text is extracted and shown to you, with a character count and a rough token estimate.
- Choose your provider and paste your API key. The key is kept in this browser's local storage and sent to nobody but the provider it belongs to.
- Pick a length - roughly 120 words for brief, 350 for standard, 900 with a section per topic for detailed - and a style: bullets, paragraphs, or a nested outline.
- Set the output language if you want something other than the document's own, and narrow the page range if only part of it matters.
- Press Summarise. The answer streams in as the model writes it, and Cancel stops the request.
Long documents are summarised in two passes. The text is cut into chunks of about 60,000 characters, each chunk is summarised on its own, and a final request writes one summary from those notes. The alternative most tools take - truncate to fit and summarise whatever survived - produces a confident account of the first thirty pages that reads exactly like an account of the whole document. This one costs more requests and reports how many it made.
There is a ceiling on how much text goes out, 120,000 characters by default, and it exists to stop someone accidentally posting a 900-page book along with the bill that would come back. When a document exceeds it, the text is cut there and the summary file opens with a line saying so - in the output you keep, not in a notification that disappears. Raise the limit under advanced options if you would rather pay for the whole thing.
The instruction the model receives is deliberately narrow: use only what the document says, note where it is unclear rather than filling the gap, and skip the preamble about being an assistant. Temperature is fixed at 0.2, low enough that the same document twice gives close to the same summary. None of that makes a summary trustworthy by itself, so treat it as a way into the document rather than a replacement for the paragraph you are about to quote.
This is one of the few pages here with an amber badge instead of a green one, because something does leave your device. To be exact: the file does not, the images in it do not, and the extracted text does - posted by your own browser, straight to the endpoint you named, with your key in the header. Nothing of ours sits between the two, so there is nothing on this side that could log it or train on it.
What this tool cannot do
- The document's text is sent to the AI provider you choose. The file itself is not, but its words are.
- You need your own API key, and your provider bills you for every request the tool makes.
- Long documents are summarised in chunks and then summarised again, so a detail mentioned once in passing may not reach the final summary.
- Tables and columns are flattened into reading order when the text is extracted, so a summary of a document that is mostly tables will be weaker than one of a document that is mostly prose.
Questions people ask
- Does my document get uploaded?
- The file does not. Its text does, and only to the provider you chose. Extraction happens in your browser, the panel shows the text and its size before anything is sent, and the request then goes from your browser directly to your provider's endpoint using your key. CekPDF has no server in that path and cannot see, log or keep any of it.
- Why do I have to bring my own API key?
- Because there is no CekPDF server, and a service that summarised documents for you would need one - along with an account, a quota, and a copy of your text passing through it. Bringing a key keeps the arrangement honest: your browser talks to your provider and nobody else is involved. The key lives in this browser's local storage, shared across the AI tools so you enter it once.
- What does it cost?
- Nothing to CekPDF - we are not in the payment path and never see your key. Your provider bills you for the tokens in the request and in the answer, and the panel estimates the size before you send. AI runs do count against your CekPDF allowance like any other task, which is what the counter on the page is showing.
- How does it handle a 300-page document?
- By cutting the text into chunks of about 60,000 characters, summarising each one, and then writing a single summary from those notes. The number of requests is reported with the result. If the document is longer than the character limit, 120,000 by default, the excess is not sent and the summary begins with a line telling you so rather than implying it covered everything.
- Can I trust the summary?
- Treat it as a fast first reading, not as a citation. The model is told to use only what the document says and to flag what is unclear, and the low temperature means the same document gives close to the same answer twice. It can still misread a table or give a passing remark more weight than it deserves. If a sentence in the summary matters, find it in the document before relying on it.
- It says my PDF has no text. What now?
- The document is a scan - a picture of a page, with no words in it to extract. Run it through OCR a PDF first, which recognises the text and adds it back as a searchable layer, then bring the result here. The check happens before any request is made, so a scan costs you nothing to discover.