Sightengine

Sightengine allows multiple detection models to be combined in a single API request. Developers can stack models such as nudity, violence, text moderation, and AI detection simultaneously, reducing latency and operation counts compared to running them separately.

Sightengine provides official client libraries for Python, PHP, and Node.js. The API also supports direct REST calls via cURL, making it compatible with virtually any programming language that can send HTTP requests.

Sightengine uses proprietary deep learning models trained to identify artifacts and patterns typical of AI-generated content. It supports detection across generators like Stable Diffusion, Midjourney, and DALL-E, returning confidence scores indicating the likelihood of synthetic origin.

Yes, live-stream moderation is available on the Pro plan and above. Sightengine processes video frames at configurable rates between 0.5fps and 2fps on Pro plans, analyzing each frame against selected moderation models and returning results in real time.

An operation is Sightengine's usage unit. Running one detection model on a single image counts as one operation. Combining multiple models in one call may consume additional operations. Failed requests and errors are not counted toward the monthly quota.