All articles

Is YOLO Free for Commercial Use? YOLOv8, YOLO11 and YOLO26 Licenses

Xuban

No. The popular YOLO models, YOLOv5, YOLOv8, YOLO11 and the new YOLO26, are not free for closed-source commercial use. They ship under AGPL-3.0, a strong copyleft license. If you build a product on one of them, you must open-source your entire application under AGPL-3.0, or buy the vendor's paid Enterprise License. If neither works for you, there is a permissively licensed alternative: LibreYOLO is MIT-licensed and free for any commercial use, no strings attached.

If you searched "is YOLOv8 free for commercial use," "YOLO license," or "YOLO without AGPL," this page answers it precisely, version by version.

YOLO licenses at a glance

ModelDefault licenseFree for closed-source commercial use?What you have to do
YOLOv5AGPL-3.0NoOpen-source your whole app, or buy an Enterprise License
YOLOv8AGPL-3.0NoOpen-source your whole app, or buy an Enterprise License
YOLO11AGPL-3.0NoOpen-source your whole app, or buy an Enterprise License
YOLO26AGPL-3.0NoOpen-source your whole app, or buy an Enterprise License
LibreYOLOMITYesNothing. Ship it in proprietary products, free.

The one-line version: the mainstream YOLO models are open source, but under AGPL-3.0, a copyleft license most companies cannot comply with in a proprietary product. LibreYOLO gives you the same YOLO workflow under the MIT license, which places no such obligation on you.

What license is YOLO under?

The widely used YOLO models, YOLOv5, YOLOv8, YOLO11, and the newest YOLO26, are distributed by their vendor under the GNU Affero General Public License v3.0 (AGPL-3.0).

AGPL-3.0 is strong copyleft. It is not the permissive "do what you want" kind of open source like MIT or Apache-2.0. It carries one specific, far-reaching obligation, and that obligation is exactly what trips up commercial users.

Is YOLOv8 free for commercial use?

Not in the way most people mean. You can download YOLOv8 and use it commercially, but only if you comply with AGPL-3.0. In practice that means:

  • If you distribute, ship, or host a product that includes YOLOv8, you must release the complete source code of that product, your app, your integration code, your scripts and config, under AGPL-3.0.
  • That obligation is triggered not only by shipping software to users, but also by letting users interact with it over a network, AGPL's defining Section 13 clause. A SaaS or API backend does not escape it.

For a hobby project or academic research, that is usually fine. For a proprietary commercial product, publishing your entire codebase is almost always a non-starter. Everything here applies identically to YOLO11 and YOLO26: same license, same obligation.

What about YOLOv5, YOLO11 and YOLO26?

Same story. All of them default to AGPL-3.0. The vendor's Enterprise License explicitly spans "YOLO26, earlier YOLO versions, and any future YOLO models," which tells you the licensing is deliberate and consistent across the whole family. Picking a newer version does not get you a friendlier license.

For completeness: not every model with "YOLO" in the name is AGPL. Some community variants are GPL-3.0 or Apache-2.0, and licenses differ by author. But the releases people mean when they type "YOLOv8" or "YOLO11" are AGPL-3.0.

What AGPL-3.0 actually requires, in plain English

AGPL-3.0 extends the GPL with one crucial addition: the network-use clause. The practical checklist:

  1. Distribute the source. If you convey software that includes AGPL code, you must make the complete corresponding source of the whole combined work available, under AGPL-3.0.
  2. Network use counts. If users interact with your modified version over a network, a web app, an API, a SaaS, you must offer them that same source. There is no "we only run it on our servers" loophole.
  3. Copyleft is viral. The obligation reaches the entire derivative work, not just the YOLO files. Your business logic gets pulled into AGPL scope.

This is general information, not legal advice, so talk to a lawyer for your exact case. But the takeaway is simple: AGPL-3.0 and closed-source commercial software do not mix.

The Enterprise license option, and its catch

The vendor offers a paid Enterprise License that removes the AGPL obligations, letting you embed YOLO in a proprietary product without open-sourcing anything. It is a legitimate path, and for some companies it is the right one.

The catches:

  • It costs money, an ongoing commercial fee, negotiated per company, with non-public pricing.
  • It is a dependency on one vendor's terms, which can change, and which cover only that vendor's code.
  • You are paying for permission to use a model architecture that, at its core, is published research.

If a recurring fee and vendor lock-in are acceptable, the Enterprise License works. If you would rather not pay, or not carry the obligation at all, keep reading.

The MIT alternative: LibreYOLO

Disclosure: LibreYOLO is ours. It is also the reason this page can end with a clean answer instead of a shrug.

LibreYOLO is a YOLO framework released under the permissive MIT license. That single fact changes everything about commercial use:

  • Free for commercial use, including proprietary, closed-source products.
  • No copyleft, no network clause. You never have to open-source your app.
  • No per-seat or enterprise fee. MIT means MIT.
  • The same workflow you already know, so migrating is straightforward.

LibreYOLO is a real, maintained framework, not a wrapper: object detection, segmentation, pose, classification, depth and more, under one familiar API, with training and ONNX/TensorRT/OpenVINO/NCNN export built in. The difference is not the workflow. The difference is the license.

If you are here from the broader "leaving the AGPL YOLO" question, we wrote a fuller comparison of the ecosystem in Best Ultralytics Alternatives in 2026.

Why MIT matters for a business

The MIT license lets you use, modify, embed, and sell software built on LibreYOLO with no obligation to disclose your source and no fee. It is the license behind much of the modern software stack precisely because it is safe for commercial adoption. You own your product; you owe nothing.

Try it

pip install libreyolo
from libreyolo import LibreYOLO

model = LibreYOLO("LibreYOLO9c.pt")
results = model("image.jpg", save=True)
print(results[0].boxes.xyxy)
print(results[0].boxes.conf)

You keep the YOLO development experience: train on your data, predict, and export to ship. You drop the AGPL obligation and the enterprise invoice.

LibreYOLO is MIT-licensed, runs on Linux, Mac, and Windows, and works on GPU, Apple Silicon, and plain CPU with no code change. One API spans YOLO9, RF-DETR, RTMDet, YOLOX, D-FINE, the RT-DETR line, segmentation, pose, depth, and more, all permissively licensed.

Star it on GitHub: github.com/LibreYOLO/libreyolo | Docs: libreyolo.com/docs