Ultralytics YOLO11 models

Ultralytics YOLO11 is a family of real-time computer vision models built on the Ultralytics open-source library. It supports object detection, instance segmentation, image classification, pose estimation, and oriented object detection with a single consistent API across training, validation, and deployment. YOLO11 improves accuracy and efficiency over earlier generations while staying easy to fine-tune on custom datasets. See the YOLO11 documentation for architecture details, benchmarks, and usage examples.

YOLO11 models are available for five tasks: object detection, instance segmentation, image classification, pose estimation, and oriented object detection. Every task shares the same training, validation, and prediction workflow.

Start with YOLO11n when latency, memory, and model size matter most. YOLO11s and YOLO11m balance speed and accuracy for most production workloads, while YOLO11l and YOLO11x prioritize accuracy when more compute is available. The YOLO11 docs include benchmarks for every size.

Use the Ultralytics library's train mode, or open the official YOLO11 project on Ultralytics Platform. After training, export your model to formats such as ONNX, TensorRT, CoreML, or LiteRT.

YOLO11 remains a supported and widely deployed model family. Ultralytics YOLO26 is the latest generation and the recommended starting point for new projects. Both families share the same API, so moving from YOLO11 to YOLO26 requires only a model change.