Pyyan / Compare / RF-DETR vs YOLO26 vs RTMDet vs Grounding DINO

RF-DETR vs YOLO26 vs RTMDet vs Grounding DINO

4 of 5

Object Detection · verified 13 Aug 2026

×RF-DETRRoboflowcurrent
×YOLO26Ultralyticscurrent
×RTMDetOpenMMLabcurrent
×IGrounding DINOIDEA Researchcurrent
1 slot left
SpecificationRF-DETRYOLO26RTMDetGrounding DINO
SummaryFirst 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.Detects whatever you describe in words.
COCO mAP60.5~55~52~52 zero-shot
SpeedReal timeVery high300+Moderate
LicenceApache 2.0AGPL-3.0MITApache 2.0
FamilyTransformer, set predictionSingle stageSingle stageOpen vocabulary
Open vocabularyNoNoNoYes
Released2025202620222023
CategoryObject DetectionObject DetectionObject DetectionObject Detection
OfficialRoboflowUltralyticsOpenMMLabIDEA Research

Highlighted rows are where these differ.

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 →

Grounding DINO

  • Text prompt in, boxes out
  • Fine-tuning still beats it on specialised objects

Best for classes you have no data for.

Full spec sheet →