Pyyan / Compare / OWLv2 vs RF-DETR vs YOLO26 vs RTMDet

OWLv2 vs RF-DETR vs YOLO26 vs RTMDet

4 of 5

Object Detection · verified 13 Aug 2026

×OWLv2Googlecurrent
×RF-DETRRoboflowcurrent
×YOLO26Ultralyticscurrent
×RTMDetOpenMMLabcurrent
1 slot left
SpecificationOWLv2RF-DETRYOLO26RTMDet
SummaryOpen-vocabulary detection trained with self-labelling.First real-time detector past 60 mAP on COCO.The latest in the line most people mean by object detection.Pure throughput, and an MIT licence.
COCO mAP~45 zero-shot60.5~55~52
SpeedModerateReal timeVery high300+
LicenceApache 2.0Apache 2.0AGPL-3.0MIT
FamilyOpen vocabularyTransformer, set predictionSingle stageSingle stage
Open vocabularyYesNoNoNo
Released2023202520262022
CategoryObject DetectionObject DetectionObject DetectionObject Detection
OfficialGoogleRoboflowUltralyticsOpenMMLab

Highlighted rows are where these differ.

OWLv2

  • Accepts an image as the query, not just text

Best for querying by example image.

Full spec sheet →

RF-DETR

  • Leads the RF100-VL domain-transfer benchmark
  • DINOv2 backbone, markedly better on occluded objects
  • Apache 2.0, which YOLO is not

Best for most custom detection tasks.

Full spec sheet →

YOLO26

  • Strong on Jetson, Snapdragon and ARM CPUs
  • AGPL-3.0: open-source your derivative work or buy a licence

Best for edge and mobile deployment.

Full spec sheet →

RTMDet

  • Wins on raw speed where accuracy is sufficient
  • MIT, so no derivative-work obligation

Best for high frame-rate video.

Full spec sheet →