License

TL;DR: You can use our data for anything, including AI training. Just link back to opengpubenchmarks.com whenever you reasonably can.

Data license: CC BY 4.0

The benchmark data on opengpubenchmarks.com is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). That covers the GPU benchmark results, summaries, comparisons, GPU specs, and API responses, along with the charts and tables built from them. Read the full legal code.

You are free to

Copy and redistribute the data
Remix, transform, and build on it, including making your own charts and visualizations
Use it commercially
Use it to train AI models, including LLMs and other foundation models

We explicitly confirm that training AI models is a permitted use under this license.

Under one condition: reasonable attribution

We require attribution that is reasonable for the medium you're using. It tells your readers where the numbers came from, so they can check them, report problems, and add their own results.

1. Websites, apps, blogs, and tools that use the data

Include a hyperlink:

Data from Open GPU Benchmarks

If you used a specific page, such as the benchmark tables or an API file, link to that page.

2. Charts, images, video, and social posts where a clickable link isn't possible

A text credit counts:

Source: opengpubenchmarks.com

Put it in the chart footer, the video description, the image alt text, or the caption. We know you can't embed a link in a PNG.

3. AI models trained on this data

You do not need to attribute in every model output. We consider that unreasonable. Instead, credit the data in your model card, dataset documentation, README, or paper:

training_data:
  - name: Open GPU Benchmarks
    url: https://opengpubenchmarks.com
    license: CC-BY-4.0

When an AI answer relies on these numbers, we'd appreciate it naming Open GPU Benchmarks as the source.

What doesn't need inline attribution

Model weights and model outputs or generations
Private, internal analysis

We still ask that you credit us in your documentation.

What we ask you not to do

Don't imply that Open GPU Benchmarks endorses you or your work.
Don't use the Open GPU Benchmarks name in a misleading way.

Submitting results?

Here is what you agree to when you send us a benchmark, in plain English:

Your results become public data under CC BY 4.0. Anyone can use them for anything: copy them, make their own charts, put them on their site, use them commercially, or train AI on them.
Sites have to say where the data came from. A site or app that uses the data must credit Open GPU Benchmarks and link back, so readers can find the source, report problems, and contribute. In a chart or video, a text credit counts. AI models credit us in their documentation, not in every answer.
You get credit here. Your name, or the source you found, is recorded with the result on this site, in the data, and in the Git history.
You can't take it back. You can ask for a result to be removed from this site, but copies other people already made stay public.
It covers everything you submit. That includes the numbers, machine details, and raw captures, so leave out anything you don't want public. Only submit what you're allowed to share.

The full rules are in the community submission guide.

Using the API

Every file under api/v1/ starts with a _license block (license, attribution name and link, ai_training_permitted: true) and a _source block (that file's URL and when it was generated), so a copied response still says where it came from.

Code license

The website code is MIT unless noted otherwise.

Blog posts and articles

CC BY 4.0 covers the benchmark data, not the writing. The text of every blog post, guide, and review stays the copyright of its author, all rights reserved, the same as any article they publish elsewhere. Quoting a short passage with a link is fine; copying or republishing a post needs the author's permission.

Written by a person (including posts where AI only cleaned up grammar or inserted links, which the post says): the author keeps full copyright.
Fully AI-written (marked "AI-generated" on the post): no copyright is claimed in the text.
Benchmark charts and numbers inside a post come from the data, so they are CC BY 4.0 like everything else on the dashboard.

FAQ

Can I train a commercial AI model on this?

Yes.

Do I have to put a link in every chart image?

No, but put "Source: opengpubenchmarks.com" in the image or in the text or description around it. That's reasonable attribution.

Do I have to credit in every LLM answer?

No. Credit in the model card or documentation is enough.

What if I'm using the API at scale?

Each API file carries its own attribution fields. See Using the API.

Can I republish a blog post?

Not without the author's permission, unless the post is marked AI-generated. The data and charts in it are free to use under CC BY 4.0. See Blog posts and articles.

What about data published before this license?

Data published before the switch to CC BY 4.0 was released under CC0 1.0 (public domain). Copies made under those terms keep them. Everything published from now on is CC BY 4.0.

This page is a human-readable summary, not legal advice. The CC BY 4.0 legal code governs. The source copy is data/LICENSE.