#7070th of 70 in Open models
yolov13
YOLOv13 real-time object detection models built on the Ultralytics codebase
- Stars
- 1.7k
- License
- AGPL-3.0
- Last commit
- Nov 2025
- Last release
- Jun 2025
- Language
- Python
1,625 on 21 Apr 20261,713 on 10 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
YOLOv13 is a research implementation of a real-time object detector, forked from Ultralytics, with Nano, Small, Large and X-Large variants. It adds hypergraph-based feature correlation (HyperACE), a full-pipeline feature distribution scheme (FullPAD) and depthwise-separable blocks. Training, validation, prediction and export to ONNX or TensorRT go through the Ultralytics YOLO Python API, with pretrained COCO weights on the releases page.
Who it is for: Computer vision engineers evaluating or fine-tuning real-time object detectors
Strengths
- Reported COCO AP50:95 of 41.6 (N), 48.0 (S), 53.4 (L), 54.8 (X), above YOLOv12 at each size
- Pretrained weights for four sizes, usable through the familiar Ultralytics YOLO API
- Exports to ONNX and TensorRT; community ports to Ascend, RKNN and Android are linked
- Includes a FastAPI REST example and public demos on Hugging Face Spaces and HyperAI
Weaknesses
- Latency in the README table is higher than YOLOv12 at every size, e.g. 14.67 ms vs 12.46 ms for X
- AGPL-3.0 license restricts closed-source and network-service use
- Install instructions pin a Flash Attention wheel for Python 3.11, CUDA 11 and Linux
- Last release was 2025-06-22; the README gives no versioned release notes
What it needs
- GPU optional
- Docker
- Needs Ultralytics, PyTorch, Flash Attention, Python 3.11
- Models: YOLOv13 N/S/L/X .pt weights, ONNX export, TensorRT export
Quick answers
Is yolov13 open source?
Yes. Its code is on GitHub under the AGPL-3.0 license.
Can you run yolov13 with Docker?
Yes. Its repository has a Docker setup: a Dockerfile, a Compose file or a published image.
Does yolov13 need a GPU?
No. It runs on a CPU and can use a GPU when you have one.
Also in Open models
See all 70| Rank | Project | Score |
|---|---|---|
| 1 | ruviewWiFi CSI sensing on ESP32 for presence, breathing, heart rate and pose | 94 out of 100 |
| 2 | layaSingle-pass multilingual classifier for typed choice, score and yes/no decisions | 88 out of 100 |
| 3 | rf-detrReal-time transformer models for object detection, segmentation and keypoints | 79 out of 100 |
| #4 | timesfmPretrained time-series foundation model for zero-shot forecasting from Google Research | 78 out of 100 |
| #5 | yolov5PyTorch YOLOv5 object detection, segmentation and classification models with training code | 76 out of 100 |