A greener city
Small changes can shape the places we live.
Trees provide shade and cool our streets.
Plants help rain soak into the soil.
A little clarity goes a long way
Similar words tell part of the story. Compare assignments side by side, explore possible paraphrases, and make room for your own judgment.
Free to use No sign-up Text stays on your device
Small changes can shape the places we live.
Trees provide shade and cool our streets.
Plants help rain soak into the soil.
Thoughtful planting makes a difference.
Trees provide shade and cool our streets.
Vegetation lets rainfall enter the ground.
Try the highlights. The real review stays in your hands.
YOUR COMPARISON SPACE
Bring your text. We’ll help you see it together.
Save a private review file to continue later. Nothing is saved automatically.
STEP 1
Class import replaces the current inputs after every file is read successfully. Draft comparison uses A as the old draft and B as the new draft.
assignment-a.txt
0 words
assignment-b.txt
0 words
Up to 1,500 words / 12,000 characters per assignment. PDF & DOCX: 5 MB. PDF: 20 pages. Scanned pages need OCR elsewhere.
Add up to two textbook excerpts or reference documents. Each assignment is compared with each source separately. These are supplied materials, not internet search results. Sources use the same size limits as assignments.
0 of 2 provided sources
Source matches do not change student-to-student coverage. Quoted or cited material can legitimately overlap. Sources are kept but not used in draft mode.
STEP 2
Start with the defaults. Expand advanced settings to change sensitivity or exclude shared material.
References: from a standalone References, Bibliography or Works Cited heading to the end. Quotes: matched straight/curly double quotes on one line. Single quotes and multiline quotations are not inferred. Review the grey exclusions; these rules can miss or exclude the wrong text.
First use takes longer while the model loads. You can cancel at any time.
STEP 3
Your results will appear here after comparison.
Each pair has its own notes and dismissed matches. Ranking changes with review decisions; it does not identify who copied whom.
Darker cells mean more highlighted words. Click a percentage to review that pair. This is coverage, not confidence or a verdict.
Ranked by the percentage of assignment words highlighted against each source. These comparisons have separate notes and statuses and are excluded from the student heatmap. Check quotations, citations and assignment instructions before drawing conclusions.
Word alignment shows edits, not writing quality. Green = added; red = removed. Moved text may appear as removal/addition. Punctuation changes are shown but not counted as words. Similarity and exclusion settings are not used.
EVIDENCE, IN CONTEXT
Coverage measures highlighted words, not plagiarism probability. Dismissed matches and excluded prompts do not count. Filters only change what is visible. Review quotations and shared topics in context.
Comparing a student assignment with a provided source. Source overlap is not added to student-to-student percentages and does not establish uncited use.
Unmarked words align in both passages. Red = wording removed from A; green = wording added in B. This alignment explains word edits, not whether the meanings are equivalent.
Notes belong to the selected pair, stay in this session and are included in its downloaded report. The highlighted PDF contains the selected pair’s extracted documents, not original PDF page images. Save review before refreshing or starting a new comparison to keep a resumable copy. Reports contain assignment excerpts.
Compare two to five English TXT, PDF or DOCX assignments, with up to 120 eligible meaning passages per pair and 300 across assignments and provided sources. Add up to two sources; their overlap remains separate from student pairs. It compares only the supplied text, uses a small quantized MiniLM model, and can miss rewrites or suggest unrelated or opposite claims. Its scores may differ slightly from the Python version. Borderline matching and simple mismatch checks reduce some false suggestions but can reject valid paraphrases.
Exact wording works without loading AI. Cancelling stops the worker; the next AI comparison may need to reload the cached model. Assignment text is not saved automatically in browser storage. Refreshing clears it; only public model files may remain cached. Save review explicitly downloads a private file with the full text, notes and decisions. Open review restores that file without rerunning AI; saved suggestions are not independently verified. Browser support and memory limits vary.
PDF import reads text and preserves physical page numbers. Check reading order for columns and tables; scanned pages and image text need OCR elsewhere. Word import reads body paragraphs and tables; headers, footers, notes, automatic numbering, images, equations and page layout are omitted. Tracked changes and text boxes may require conversion first.
Current desktop Edge and Chrome support the tested workflow. Other browsers and smaller devices still need compatibility testing. This tool compares only supplied assignments; it does not search the internet or detect AI authorship. Open-source dependencies and licenses.
LESS GUESSWORK. MORE CONTEXT.
Go from a pile of assignments to a clearer conversation.
Paste text or choose TXT, PDF and DOCX files. Compare two assignments, a class of up to five, or two drafts.
Find shared wording and possible paraphrases. Add provided sources or exclude common assignment questions.
Read passages in context, keep notes, dismiss suggestions and export a report. Similarity is evidence to review, not a verdict.
YOUR TEXT STAYS YOURS
Assignment text is processed in your browser. No account, no assignment uploads, no paid API.
On first use, possible paraphrases loads about 39 MB of runtime and model files. Your browser runs the model on its CPU and may cache it for later visits. Speed depends on your device. Exact wording works without loading AI.
Only your theme preference is remembered in this browser. Assignments and notes are not saved automatically. Refreshing clears your work. Save review explicitly downloads a private file containing text, notes and decisions; keep that file somewhere you trust.
Highlights show overlap in the supplied material. They do not establish plagiarism, copying direction or AI authorship. Possible paraphrases can be wrong or missed. Review citations, assignment instructions and the original meaning.