Educators
Use signals to guide a conversation about how a student approached an assignment.
Add one snippet and choose its language, or leave the setting on Auto.
The detector looks at style, naming, structure, and other patterns in the code.
Review the likely label, the model’s confidence, and the signals behind it.
88% accuracy for Flash on the tested subset and 96% for Sol on the full blind test.
The blind test used 100 snippets, pairing human-written and AI-generated solutions to the same problems and hiding their labels during scoring.
Use signals to guide a conversation about how a student approached an assignment.
Identify parts of an interview submission or outsourced work to discuss with the author.
Use the estimate as one input when deciding which sections need a closer review.
Check a snippet for signs of AI assistance and review the signals before acting.
AI code checkerStart with free checks. Pro is listed at $9.99/month or $99/year.
See the blind test and its measured error rates on the accuracy page. View methodology
This result is a probability estimate, not proof. Never use it as the sole basis for academic or employment decisions. Detection is probabilistic — always review the signals yourself before acting.
It is the model’s confidence in the label it selected. Below 66% is low, from 66% to 74% is medium, and 75% or higher is high. The score is an estimate for the snippet you submitted.
We send it to APIMart for analysis and store detection metadata such as the language, estimated label, confidence, and time. We remove submitted code after seven days and never execute it or use it for training.
You can run 5 checks per day without an account. A free account gives you 10 checks per day.