GUIDES · 2026
If you write test papers for a school or college, you already know the tedious part isn't coming up with content — it's turning your notes, a textbook chapter, or a slide into a formatted, printable question paper with an answer key. Our free Auto MCQ / Test Paper Generator automates that last step: photograph your source material, let it extract the text, review it, and get a ready-to-print MCQ paper in a few minutes — entirely in your browser, with no paid AI subscription required.
Start with whatever you already have — a page of handwritten or printed notes, a textbook chapter, a PDF slide, or a photo of a whiteboard. A clear, well-lit, straight-on photo works best; avoid heavy glare, extreme angles, or blurry handwriting, since OCR (optical character recognition) reads printed and typed text far more reliably than cursive handwriting.
Upload the image(s) on the MCQ Generator page — you can select or drag & drop multiple photos at once, for example every page of a chapter, and remove any you don't want with the small "×" on its thumbnail. Click "Extract Text From All Images" and it runs OCR on each image in turn, showing combined progress across the whole batch (e.g. "Processing image 2 of 4…"), and appends each image's text into the same box separated by a clear "--- Image N ---" divider so you can see where each page's content starts. If one image fails to process, the tool notes it and keeps going with the rest instead of stopping the whole batch. It all runs on Tesseract.js, an open-source OCR engine that runs via WebAssembly directly in your browser — your photos are never uploaded to a server. The first extraction in a language you haven't used yet may take a few extra seconds while a small language data file downloads and gets cached; after that, it's fast.
If you don't have notes to photograph — say you just need a quick test paper on "Photosynthesis" or "Newton's Laws of Motion" — switch the tool's mode tab to 🔎 From Topic (Auto-Research), type in the topic, and click "Search & Fetch Content." The tool looks up that topic on Wikipedia's free, public, keyless API and pulls in a plain-text article, which lands in the exact same editable text box the OCR path fills. It's worth being clear about what this is and isn't: it searches Wikipedia specifically, not the whole internet, and it doesn't call any paid AI model — it's a straightforward, free lookup. From there, everything works exactly the same as the OCR flow: proofread the fetched text, then generate your MCQs from it.
OCR is good, not perfect. Numbers, similar-looking letters, and unusual formatting can get misread, especially from photographs rather than clean scans. The extracted text lands in an editable box specifically so you can fix mistakes before generating any questions — a minute spent proofreading here saves you from a test paper with garbled blanks or nonsense answer options.
Add your school or institution name, then choose Subject and Class/Grade from the dropdowns (each has an "Other…" option with a text field if your subject or class isn't listed). Picking a Class fills in a note that suggested defaults are available for MCQ count, short-question count, long-question count, total marks, and time allowed, tuned to a rough class tier — primary (Class 1–5), middle (Class 6–8), matric (Class 9–10), or intermediate (Class 11–12). Be clear about what these are: Pakistan doesn't have one single, universally-mandated exam-board pattern, so these are sensible, editable starting suggestions, not an official rule. The tool never silently overwrites a field once you've typed into it yourself — if you want to reapply the suggestion anyway, use "Reset to Suggested Pattern." Round out the rest of the form with chapter/topic, test title, date, and (optionally) your name as the teacher and any instructions for students.
Set how many MCQs (up to 30), short questions (up to 15), and long questions (up to 10) you want — each is independent, and 0 is a valid value if you only want one or two of the three types. Then click "Generate MCQs."
There's no paid AI API behind this — it's a transparent, rule-based algorithm that runs locally, and it's honest about what it is: it does not write original explanations. It splits your text into sentences, filters out ones that are too short or too long, and looks for sentences containing a meaningful "key term" — a proper noun, a number or date, or a longer, less-common word.
For MCQs, it blanks out that term to form the question stem, then builds four answer options by pulling other real terms from your own source text wherever possible (falling back to generic distractors only when the text doesn't have enough unique terms), and shuffles the option order so the correct answer isn't always in the same position. For short questions, it picks a different, not-already-used sentence and wraps its key term in a natural prompt like "What do you understand by '…'? Explain in 1–2 sentences." — the answer key shows your own original sentence as the "Reference answer (from source text)," since the tool is reformatting your content, not composing a new explanation. For long questions, it groups a few related, not-already-used sentences into a passage, and phrases the prompt as a general essay instruction keyed on that passage's dominant term (e.g. "Write a detailed note on '…'. Include relevant facts, causes, or examples where applicable.") without revealing the passage itself in the question — the answer key shows the joined source sentences as the reference answer. If your text doesn't have enough distinct candidate sentences to avoid overlap between MCQs, short and long questions, the tool reuses content as a fallback and tells you exactly what it generated for each type instead of failing silently.
The preview shows a full letterhead-style paper — your institution name, subject/class/title line, date/time/marks row, a Name and Roll No. line for students, your instructions, and clearly separated, numbered sections: Section A (MCQs, with all four A/B/C/D options laid out on one line), Section B (Short Questions), and Section C (Long Questions) — whichever sections you requested, sharing one running question number. A live "Estimated pages" counter and a "Fit to page(s)" dropdown let you shrink the paper to fit 1, 2, or 3 printed pages before you print. Click "Print / Save as PDF" to export the clean test paper (the site header, form, and everything else is hidden automatically in print), and use "Print Answer Key Only" to print a separate answer key for yourself — with the MCQ letter-answer grid plus labelled reference answers for short and long questions — it never shows up on the student copy.
Questions come from a transparent, rule-based algorithm — not a metered AI API.
OCR and question generation both run locally — your notes are never uploaded anywhere.
A clean letterhead paper plus a separate answer key, ready for "Save as PDF."
Ready to try it? Head over to the Auto MCQ / Test Paper Generator and turn your next set of notes into a finished test paper in a few minutes.