API Changelog
Track major additions and changes to the Sightengine APIs. This page lists user-facing updates to our APIs. Internal improvements to accuracy, latency, and performance happen frequently and are not documented here.
May 2026
- AI Image & Video Detection — Major improvement: support for AI edit detection (image-to-image edits).
major upgrade
April 2026
- Image Recapture Detection — Release of a new model recapture to evaluate whether a photo is original, or if it is a photo of a printed or displayed image.
- AI Image & Video Detection — New generator class kling for the Kling generator family. This family was previously returned in the other class.
new product
new class
March 2026
- Face Liveness detection — New model liveness, to evaluate whether a face is physically present at the moment of capture or if it is a spoofing attempt using screens, prints, masks, or synthetic media.
- Face Presence Detection — New subclasses for out-of-scene faces, to help distinguish between faces present through embedded overlays, and faces present through screens, posters, photos.
- Text moderation & OCR moderation — Romanian: New language added to rule-based text moderation and ocr moderation in images & videos: romanian. ISO-code: ro.
new product
new class
new feature
February 2026
- AI Image & Video Detection — New generator classes higgsfield and seedream. The Higgsfield and Seedream generator families were previously returned in the other class.
- OCR moderation — New paranoid mode to maximize recall for phone number detection. This mode is activated through the opt_phone parameter. See details.
new class
new feature
January 2026
- OCR moderation — New major version introduced through the text-content-2.0 model.
- AI Image & Video Detection — The generator class gpt4o is renamed to gpt.
- AI Image & Video Detection — New generator class z_image for the Z-image generator family. The Z-image generator family was previously returned in the other class.
major upgrade
other
new class
November 2025
- AI Music Detection — New model to help detect AI-generated music.
new product
October 2025
- AI Image & Video Detection — New generator classes qwen and wan. The Qwen and Wan generator families were previously returned in the other class.
new class
September 2025
- Image & Video Search — Image & Video similiarity upgrade. The similarity evaluation engine has been upgraded to its 3rd iteration, providing major accuracy improvements on adversarial cases with strong image modifications. All image and video lists created after September 2025 automatically use this new version. Previously created lists continue running using the older similarity evaluation engine.
major upgrade
July 2025
- Face Presence Detection — New model face-presence to indicate whether a detected face belongs to a person who is physically present within the photographed scene.
new product
June 2025
- People counting — New model people-counting to detect and analyze the presence of people in images and videos.
new product
May 2025
- Face age — New model face-age to estimate the age group of people appearing in photographs, based on face analysis.
new product
April 2025
- Text moderation & OCR moderation — Korean: New language added to rule-based text moderation and ocr moderation in images & videos: korean. ISO-code: ko.
- AI Image & Video Detection — New generator classes gpt4o and reve. Those generators were previously returned under the other class.
new feature
new class
March 2025
- Face analysis — New model face-analysis to analyze faces in images and videos and provide insights on realness, position, obstruction, quality, angle, and filters. This new model introduces several new concepts and classes: obstruction quality angle filters and helps distinguish real faces from artificial ones.
- Text moderation & OCR moderation — Russian: New language added to rule-based text moderation and ocr moderation in images & videos: russian. ISO-code: ru.
new product
new feature