AI Flashcard Generator — turn notes into study cards
Paste notes, a passage, or a glossary and get a deck of flashcards in seconds — question/answer, term/definition, or fill-in-the-blank. Study them with spaced repetition and export to Anki, Quizlet, CSV, or JSON. No signup, and study and export work offline.
How it works
The generator turns free text into cards with a two-track pipeline. The deterministic track is always available — it runs in your browser, needs no network, and produces the same deck every time — while an optional neural track adds question/answer cards when the server has a Hugging Face key configured. Nothing you paste is stored.
- Segment. Your text is whitespace-normalised and split into sentences with a Unicode-safe splitter that respects
. ? !and skips common abbreviations (Dr., e.g., etc.). Empty and duplicate sentences are dropped. - Term–definition and Q&A (deterministic). A copula parser finds patterns like “X is Y”, “X refers to Y”, and “X means Y”. The subject becomes the front (or the question “What is X?”) and the predicate becomes the back. Glossary lines written as
Term: definitionare split on the first delimiter. - Cloze (deterministic). For fill-in-the-blank cards, each token in a sentence is scored: a number scores highest, then a capitalised proper noun, then the longest content word. The single top-scoring token is blanked with
[...]on the front and revealed on the back. Because the rule is fixed, the same sentence always yields the same card — so the cloze path can be reconciled by hand and is cross-checked by a second, independent scorer on every run. - Neural Q&A (optional). When enabled, the cleaned text is sent once to the
google/flan-t5-basemodel through the Hugging Face Inference API — the same server-side pattern the site's summarizer and question-answering tools use. Its cards are merged ahead of the deterministic ones. If the model is unavailable, the deterministic deck stands on its own; you never see a dead state. - Study & export.Study mode applies the Leitner system: every card starts in Box 1, “Got it” promotes it a box, and “Review again” resets it to Box 1; the next card is always drawn from the lowest box. Exports (Anki, Quizlet, CSV, JSON) are pure string serialisations built in your browser — the CSV is RFC-4180 quoted and the Anki/Quizlet files are tab-separated, one card per line.
Up to 8,000 characters can be processed at once. Generated cards are a study aid, not a source of truth — always sanity-check them against your material, especially for exam revision.
Worked examples
Frequently asked questions
Sources & references
- Anki Manual — Importing (tab-separated field format)
- Quizlet Help — Importing a set (term/definition delimiters)
- Leitner system — the spaced-repetition box method used in study mode
- google/flan-t5-base — the instruction-tuned model used for neural Q&A
- RFC 4180 — the CSV field-quoting spec the CSV export follows
The card-format and study-method references above were last cross-checked on 2026-07-09. The deterministic cloze and term–definition worked examples on this page are reproducible from the tool exactly.
Related tools
Comments & feedback
Spotted a bug or want an improvement? Tell us — our team reviews every comment, and good ideas get built. Comments are public and anonymous.
Found a bug, edge case, or want to suggest an improvement?
Email me at [email protected] — most fixes ship within 24 hours.