Apache 2.0
11 of the 24 models in the AI Token Router catalogue are published under Apache 2.0: Qwen3 Max Instruct, Qwen3 235B A22B, Mistral Large 3, GPT-OSS 120B, Wan 2.2 T2V A14B, Wan 2.2 I2V A14B, CogVideoX-5B, Mochi 1, FLUX.2 [schnell], Qwen-Image and Qwen3 Embedding 8B. Commercial use is permitted outright by the licence text.
Apache 2.0 is a standard permissive licence: it allows commercial use, modification and redistribution provided the notices are kept and changes are stated, and it adds an express patent grant from the contributors that terminates if you bring a patent claim over the work.
A standard permissive text that allows commercial use outright, with no user threshold and no field-of-use restriction.
This is a reference, not legal advice. It exists so you can arrive at your own counsel with a sharper question than whether you are allowed to use the model.
What the licence supports
Permits
- Commercial use, including inside a product you charge for
- Modification, including fine-tuning, and distribution of derivative works
- Patent use, under an express grant from each contributor
- Private use, with no obligation to publish anything you build
Requires
- Keep the copyright, patent, trademark and attribution notices
- State which files you changed
- Include a copy of the licence with any redistribution
Restricts
- The patent grant terminates if you bring a patent claim alleging the work infringes
- Trademarks are not licensed — the grant covers the work, not the publisher's name or marks
- No user threshold and no field-of-use restriction is imposed by the text
The canonical text is at https://www.apache.org/licenses/LICENSE-2.0. Read it rather than this summary if anything you are shipping turns on it.
Models under Apache 2.0 (11)
2 of these are callable today. Availability is a fact about our upstreams and not about the licence: a model listed here and not yet served carries exactly the same terms.
| Model | Publisher | Type | Availability |
|---|---|---|---|
| Qwen3 Max Instruct | Alibaba | text | Not served yet |
| Qwen3 235B A22B | Alibaba | text | Not served yet |
| Mistral Large 3 | Mistral AI | text | Not served yet |
| GPT-OSS 120B | OpenAI | text | Not served yet |
| Wan 2.2 T2V A14B | Alibaba | video | Not served yet |
| Wan 2.2 I2V A14B | Alibaba | video | Not served yet |
| CogVideoX-5B | Zhipu AI | video | Not served yet |
| Mochi 1 | Genmo | video | Not served yet |
| FLUX.2 [schnell] | Black Forest Labs | image | Not served yet |
| Qwen-Image | Alibaba | image | Callable |
| Qwen3 Embedding 8B | Alibaba | embedding | Callable |
Questions
Can I use Apache 2.0 models commercially?
Yes. Apache 2.0 is a standard permissive licence: it allows commercial use, modification and redistribution provided the notices are kept and changes are stated, and it adds an express patent grant from the contributors that terminates if you bring a patent claim over the work. This is a reference, not legal advice. It exists so you can arrive at your own counsel with a sharper question than whether you are allowed to use the model.
What does Apache 2.0 require me to do?
Keep the copyright, patent, trademark and attribution notices. State which files you changed. Include a copy of the licence with any redistribution. Those are the conditions the standard text attaches; the text itself is the authority and is linked on this page.
Which models under Apache 2.0 can I call today?
2 of the 11: Qwen-Image, Qwen3 Embedding 8B. The rest are catalogued at their intended rate but not currently served by an upstream. Across the whole catalogue, 7 of 24 models are callable today.
Other licence families
The full licence table lists all 24 catalogued models against their licence, and names the closed-weight models we refuse to resell. The accompanying article argues why the difference between these families decides how long a procurement review takes.