Krex AI

Generation language

Default language for AI-written scripts, TTS voiceovers, and captions. You can still override per-job.

Analyzer suite

Analyzer · Plagiarism

Your distinctive sentences, searched where they'd surface

The check selects your most distinctive sentences and searches them across the public web and the CORE open-access research index — about 300 million scholarly works. Matches come back side-by-side with similarity levels and links to the sources.

2 credits per 8,000 words · up to 16,000 words per scan

Web + open access research

Two arms in parallel: exact-phrase web search, and the CORE index of open-access papers — the scholarly overlap a plain web search misses.

Side-by-side evidence

Every match shows your passage against the source passage with the overlap highlighted and the source linked — evidence you can check, not a bare percentage.

Distinctiveness-first sampling

Common phrases match everything and prove nothing. The check searches your most identifying sentences — where real copying would actually surface.

Honest scope, stated on the page

Paywalled journal databases are contractually exclusive to legacy vendors at 10–100× our price. We search what's actually searchable and say so — a clean result is not a certificate.

How it works

01

Paste or upload

Up to 16,000 words per scan. Longer documents sample proportionally — the coverage note tells you exactly what was searched.

02

Two-arm search

Distinctive phrases go to the web and the open-access index in parallel; results are verified by local overlap scoring before you see them.

03

Review matches

Each match: your text, their text, similarity level, and the linked source — labeled web or open-access paper.

What a clean result means

A clean result means nothing was found on the public web or the open-access index. It is not a guarantee of originality — paywalled corpora and print sources are outside any affordable tool's reach, and we won't pretend otherwise.

Straight answers

Does it search paywalled journals?

No — and neither does any tool at consumer prices. The major paywalled similarity corpus is contractually exclusive to one legacy vendor. We search the public web plus ~300M open-access works and label every source.

Why sentences instead of the whole document?

Exact-phrase search on distinctive sentences is how real copying surfaces. Searching every sentence would mostly match boilerplate and inflate false positives.

What are the similarity levels?

Each match is scored by local overlap between your passage and the source. High similarity with a linked source is evidence worth reviewing; we show the passages so you can judge.

The other instruments

Free to start — 50 credits a month, no card.

Plagiarism is one of six instruments sharing one document — check, verify, and polish without pasting twice.