LIVE DEMO / CLASSICAL V2

Put an image to the test.

Upload a still image to run our actual classical detector and inspect the features behind its estimate.

Ready to analyze.

The smaller model, clearly identified.

This free demo runs the 85-feature classical v2 model. It does not run the DINOv2 embedding or the full 855-feature unified model.

Classical benchmark: 78.1% accuracy and 0.863 ROC-AUC on the same 6,414-row test split. These are evaluation results, not a guarantee for your image.

Still images only. Unfamiliar generators, real artwork, small images, and heavy processing can produce incorrect estimates. Do not use this result as proof or as sole evidence for consequential decisions.

Before you upload

Your image is sent to our demo service on Render. The application uses a temporary copy, removes it when analysis finishes, and does not use uploads for training. The hosting provider processes network requests under its own privacy policy.

Render privacy policy

This is a free demonstration with limited capacity. After inactivity, the service may take about a minute to wake up. Try again if it is busy.

The benchmark. The boundaries.

Put an image to the test.

Estimated AI probability