🎓PDF to Quiz
Drop a PDF into PDF to Quiz and get a 10-question multiple-choice quiz back with an answer key and page citations for every question. The PDF is parsed entirely in your browser, the file never uploads to our server. Extracted text with page markers is sent to Anthropic Claude, which generates plausible distractors, one correct answer per question, and a brief explanation of each correct answer pointing to the source page. The quiz is delivered as plain markdown so it pastes into Notion, Google Docs, Word, or any. LMS that accepts markdown.
Why PDF to Quiz changes test prep
Writing a 10-question multiple-choice quiz from scratch takes a teacher 30 to 45 minutes if they are working from a textbook chapter they know well. Coming up with plausible distractors is the slowest part, wrong answers that students might pick if they only half-read the material, not so obviously wrong that they give the answer away PDF to Quiz collapses this loop. Drop the PDF in, wait twenty seconds, and review a draft quiz where every question has been generated from the source text with a page citation for verification. You spend the saved time on the editing pass, swapping out questions you do not like, tightening wording, rather than building from a blank page.
The quiz format is opinionated. Ten questions because that hits the sweet spot for a single review session, long enough to cover the material, short enough to grade quickly. Exactly one correct answer per question because true/false hybrids ("two of the above") inflate guessing rates and confuse weaker students. Four options. A through. D because three is too easy and five is too many. Distractors are written to be plausible, not obviously wrong, not trick answers, so the question actually tests understanding rather than reading comprehension. An answer key with one-sentence explanations and page citations follows each quiz, which makes self-graded study sessions actually useful.
The page citation is the most important feature. When a student gets a question wrong, the citation tells them exactly where to look in the source PDF. When a teacher edits a question that does not feel quite right, the citation tells them which paragraph the AI was working from When an auditor or curriculum reviewer wants to confirm the quiz reflects the source material rather than the AI making things up, the citation provides an audit trail. Other AI quiz generators skip this step. We do not, because it is the difference between a tool you can trust for actual assessment and a toy you have to babysit.
The PDF extraction is client-side using Mozilla pdf.js, the same library that powers PDF rendering in Firefox. Your file never uploads to our server. The extracted text (with page markers) is sent to Anthropic Claude for the quiz generation step, never stored, never used to train a model. For PDFs that are scans without an OCR text layer (older textbooks, scanned worksheets), the extractor returns empty text and you will see a notice, run the file through the FixTools OCR PDF tool first to add a text layer, then come back and generate the quiz.
How to use PDF to Quiz
- 1
Upload your study PDF
Drop the file onto the upload area or click to browse. Textbook chapters, lecture notes, articles, compliance training material all work. The file stays in your browser, there is no upload to our server during extraction. A 30-page chapter typically extracts in under five seconds.
- 2
Confirm page count and any tier limit notice
You will see the page count of the PDF. If the document exceeds your tier limit (10 free / 200 paid), a notice will tell you the first. N pages will be used. Most single-chapter PDFs fit comfortably in the free tier.
- 3
Click Run PDF to Quiz
The extracted text is sent to Claude with explicit page markers. The model generates 10 multiple-choice questions, picks one correct answer per question, writes three plausible distractors per question, and produces an answer key with explanations and page citations. This step takes fifteen to twenty seconds.
- 4
Review the quiz draft
Read through the questions. Swap out any that feel weak by editing the markdown directly. Check the page citations, open the PDF, jump to the cited page, confirm the question reflects the source. This is your editing pass.
- 5
Copy or export
Use the Copy button to grab the entire quiz as markdown. Paste into Notion, Google Docs, Microsoft Word, Canvas, Moodle, or any tool that accepts markdown formatting. For. LMS exports (Quizlet, Anki, GIFT format for Moodle), that conversion is on the roadmap; in the meantime markdown copy-paste works in every major editor.
Real-world use cases
High school teacher preparing a chapter review for AP Biology
A teacher uploads chapter 12 (cellular respiration, 24 pages) on Sunday evening PDF to Quiz produces 10 questions covering glycolysis, the Krebs cycle, oxidative phosphorylation, and ATP yield with page citations. She rewrites three questions to match her teaching emphasis, prints the quiz for Monday's warm-up. Time spent: 12 minutes instead of the usual 35.
College student building a practice test from a textbook PDF
PDF to Quiz generates a 10-question quiz across cell-mediated immunity, antibodies, and immunological memory. The student self-grades against the answer key, identifies they're weakest on T-cell signaling, returns to those pages in the source for deeper review.
Corporate L&D building a quarterly compliance refresher
An L&D specialist uploads the updated data privacy policy (18 pages) and gets a 10-question quiz covering employee responsibilities, PII handling, breach reporting, and GDPR requirements with page citations. She drops the quiz into the company. LMS as a required quarterly training. The page citations let any auditor verify the quiz reflects the actual policy document.
Bootcamp instructor generating practice quizzes from JavaScript documentation
A coding instructor uploads the async/await section of an O'Reilly. JavaScript PDF (12 pages) and gets a quiz testing. Promise resolution order, error handling, parallel vs sequential awaits, and common pitfalls. He runs the quiz at the end of week 6 of the bootcamp; students who score below 7/10 get a one-on-one review session.
Pro tips
💡 Run multiple quizzes and combine the best questions
Generate two or three quizzes from the same PDF. You will get different question selections each time. Cherry-pick the strongest 10 across the runs, this gives you better coverage of the material than any single run.
💡 For final exam review, split the PDF and generate per-section quizzes
A whole-textbook-chapter quiz forces 10 questions to cover 30 pages, coverage is necessarily thin. Use FixTools PDF Splitter to break the chapter into 4-5 page sections, generate a quiz per section, combine for a 50-question chapter review with stronger coverage.
💡 Verify questions against the cited page before using the quiz for grading
For low-stakes review the AI-generated quiz is fine as-is. For graded quizzes, open the PDF to each cited page and confirm the question and answer accurately reflect the source. The citation is the audit trail, use it before grades depend on the result.
💡 Edit the distractors when they are too obvious
AI distractors are usually plausible but occasionally one is too obviously wrong (anachronism, unit error). Edit those manually, the right distractor should be something a student who skimmed the chapter would seriously consider before dismissing.
Frequently asked questions
How many questions does it generate?
10 questions by default. The system prompt is calibrated for 10 because it is the sweet spot for single-session review. If the source PDF is too short to support 10 distinct questions, the model generates as many as the material supports and notes the count at the top. To get more than 10, run the tool multiple times, you will get a different question set each time.
Is the PDF uploaded to your server?
No PDF text extraction happens entirely in your browser using Mozilla pdf.js, the same library that powers PDF rendering in Firefox. The PDF file never leaves your device. Only the extracted text (with page markers) is sent to Anthropic Claude for the quiz generation step. We never store the text, the quiz, or the source file.
Can I export to Quizlet, Anki, or Moodle?
Direct platform exports (GIFT for Moodle, .apkg for Anki, Quizlet CSV) are on the roadmap. For now the output is markdown, every major editor and. LMS accepts markdown, and conversion to platform-specific formats is a copy-paste step away. The /ai/pdf-to-flashcards tool (coming) will ship native Anki and Quizlet CSV export.
Does it work with scanned PDFs?
Only if the scan has an OCR text layer. Most modern scanners (Adobe. Scan, iOS Files app, Genius. Scan) add OCR automatically. Older textbook scans or photographed pages often have no text layer, the extractor will return empty text. Run the file through the FixTools OCR PDF tool first to add a text layer, then come back and generate the quiz.
How accurate are the answer keys?
Claude is the strongest LLM available for structured tasks like quiz generation, and the page citation on every question lets you audit the answer against the source. In our testing, answer key accuracy is well above 95% for well-structured source documents (textbook chapters, articles, policy documents). For graded quizzes, always verify a sample of questions against the cited pages before using.
Can teachers use this for graded assessments?
Yes, with an editing pass. The AI generates a solid first draft, typically 8 or 9 of the 10 questions are usable as-is, with one or two needing distractor improvements or wording tightening. Treat the output as a strong starting point, not a finished assessment. Page citations make the editing pass faster because you can verify each question against the source.
Does it support languages other than English?
Yes. pdf.js extracts text from any language a PDF can contain Claude generates the quiz in the language of the source document by default, upload a Spanish-language textbook chapter and you get a Spanish-language quiz. For mixed-language sources or to specify the output language, edit the question prompt manually.
How is this different from ChatGPT pasting the PDF text?
Three differences: (1) page citations on every question, you can verify each item against the source rather than trusting the LLM; (2) no upload to a server during extraction, PDF stays in your browser; (3) opinionated output format (10 questions, single correct answer, four options, plausible distractors, answer key with explanations), ChatGPT can do this but you have to prompt for the structure each time and the page citations are unreliable.