Ultralytics YOLO-OCR

YOLO-OCR is an upcoming Ultralytics model for optical character recognition — finding and reading text in images and video. It is announced on the official Ultralytics roadmap as fast, accurate text recognition, bringing OCR into the real-time YOLO workflow used for detection, segmentation, and pose.

Most OCR pipelines are built for scanned documents and run too slowly for live video. Real-time OCR unlocks reading in the physical world: license plates and containers in logistics, serial numbers and labels on production lines, shelf labels in retail, and signs or displays in robotics — all processed at camera frame rates.

Ultralytics announces release timing on the official roadmap, which is the authoritative source for what ships when. Follow the roadmap and Ultralytics on GitHub for release announcements.

Today, a practical pattern is training Ultralytics YOLO26 to detect text regions, labels, or plates, then passing crops to an OCR engine. The roadmap describes YOLO-OCR as fast, accurate text recognition, and building on the Ultralytics workflow now positions those pipelines to evaluate it at launch.

Watch the Ultralytics roadmap, star the GitHub repository, and create a free Ultralytics Platform account — new models land there for training and deployment when they are released.