Model weights · YOLO26-obb

Download YOLO26-obb weights

Oriented bounding boxes: each detection carries a rotation angle, so a diagonal ship, plane or pallet is enclosed tightly instead of by a box mostly full of background. Pick a size below and download the official .pt checkpoint in one click, with published COCO accuracy, size and license all in view.

AGPL-3.0Ultralytics · 2026 · Ultralytics (PyTorch)

YOLO26-obb checkpoints, pick a size and download

Oriented bounding boxes: each detection carries a rotation angle, so a diagonal ship, plane or pallet is enclosed tightly instead of by a box mostly full of background.

  • YOLO26n-obb
    yolo26n-obb.pt
    DOTAv1 mAP50
    78.9
    Params
    2.5M
    Size
    5.6 MB
  • YOLO26s-obb
    yolo26s-obb.pt
    DOTAv1 mAP50
    80.9
    Params
    9.8M
    Size
    20.7 MB
  • YOLO26m-obb
    yolo26m-obb.pt
    DOTAv1 mAP50
    81.0
    Params
    21.2M
    Size
    46.1 MB
  • YOLO26l-obb
    yolo26l-obb.pt
    DOTAv1 mAP50
    81.6
    Params
    25.6M
    Size
    54.6 MB
  • YOLO26x-obb
    yolo26x-obb.pt
    DOTAv1 mAP50
    81.7
    Params
    57.6M
    Size
    121.1 MB

Scores are DOTAv1 mAP50 at 1024px, published by the authors. Weights host: github.com. Clicking Download verifies the file and starts it straight from the official CDN.

How to load YOLO26-obb weights

Install the Ultralytics package, then point the loader at the checkpoint. It downloads automatically on first use, or you can pass the local path to the file you downloaded above.

pip install ultralytics
from ultralytics import YOLO

# Downloads on first use, or pass a local path to your .pt file
model = YOLO("yolo26n-obb.pt")
results = model("image.jpg")

Want to see the architecture? Export to ONNX and open it in the ONNX visualizer for a labelled diagram with shapes and parameter counts.

Open visualizer →

Choosing a YOLO26-obb checkpoint

Read off the table above: what each step up the size ladder costs and returns for this family specifically, where it stops being worth it, and what the license actually permits.

What the size ladder buys

YOLO26-obb spans YOLO26n-obb at 2.5M parameters to YOLO26x-obb at 57.6M, 23.0× the model, for 2.8 more DOTAv1 mAP50 (78.9 to 81.7). That works out to roughly 0.05 points per additional million parameters across the whole range, and the return is front-loaded: the early steps are much cheaper than the last one. Published figures from Ultralytics at 1024px, not re-measured here.

Where the ladder stops paying

The best value step is YOLO26n-obb → YOLO26s-obb, worth 0.27 DOTAv1 mAP50 points per million parameters. The worst is YOLO26l-obb → YOLO26x-obb at 0.00, or 88× less efficient, for 32.0M extra parameters and only 0.1 more points. If you are latency- or memory-bound, that is the step to skip; if you are accuracy-bound and the compute is free, it is the only place left to get it.

Licensing, in practice

The weights are AGPL-3.0. AGPL-3.0 is free for open-source, research and internal use; a closed-source commercial product needs an Ultralytics Enterprise license.

About YOLO26-obb

YOLO26-obb predicts a rotated rectangle per object rather than an axis-aligned one, which matters whenever the things you are detecting are not conveniently square to the camera. Overhead imagery is the classic case: a ship at 45 degrees fills roughly half of its axis-aligned box, so counts, spacing and packing all get worse the more rotated the scene is. These checkpoints are pretrained on DOTAv1 at 1024px and scored as mAP50, which is a different dataset, resolution and threshold from the COCO numbers elsewhere in this catalog; do not read them against each other.

Author
Ultralytics
Released
2026
Tasks
OBB
Framework
Ultralytics (PyTorch)
Input size
1024px
License
AGPL-3.0
Resources
DocsGitHubLicense

More model weights to download

Each page lists every checkpoint with accuracy, parameters, license and a one-click download.

See all 19 model families →

YOLO tracker FPS benchmarks →Get in touch →
Blog

From the blog

Tutorials, code, and notes on computer vision, deep learning, and applied AI.

View all posts →

Frequently asked questions

How do I download YOLO26-obb weights?
Click the Download button next to any variant in the table above; we verify the file and start it straight from the official CDN. You can also let the loader fetch it automatically on first use with `model = YOLO("yolo26n-obb.pt")`. All 5 YOLO26-obb checkpoints are hosted officially.
Is YOLO26-obb free for commercial use?
YOLO26-obb is released under AGPL-3.0. That is free for open-source and research; closed-source commercial deployments need an Ultralytics Enterprise license. Always confirm against the linked license text.
Which YOLO26-obb model size should I use?
Start with YOLO26n-obb, the smallest and fastest, ideal for prototyping, edge and CPU. Move up the ladder only when you need more accuracy and have the compute for it. The largest variant reaches 81.7 COCO mAP.