Definition
Apache 2.0 license is the Apache 2.0 license is a permissive open-source license that lets anyone use, modify and sell software or a model commercially, as long as they keep the license notice and state their changes.
Apache License 2.0 is one of the most common open-source licenses. It is published by the Apache Software Foundation and is used for software and, increasingly, for AI models. When a model is released under it, you get broad rights to use it without asking permission or paying a fee.
The conditions are light. You must include a copy of the license and any notices, and say clearly which files you changed. You may not use the original authors' names or trademarks to promote your product without permission.
It also includes an express patent grant. Contributors give you a license to any patents their contribution covers, and that license can end if you sue them over those patents. Many companies prefer Apache 2.0 for this reason.
Not every model labeled "open" is as open as this. Some release model weights under custom licenses that restrict commercial use, cap the number of users, or limit certain fields. An open-weight model under Apache 2.0 has none of those limits, so you can plan a business around it.
Two cautions:
A team wants to run a coding model inside its own network so no code leaves the building. A model under Apache 2.0 allows that, along with adapting it with LoRA and shipping the result in a product. A model under a restrictive custom license may not.
The license governs the use of open-weight models, and it is separate from whether the model is any good. Test it on your own tasks with evals before relying on it. Running it yourself also means paying for the compute it needs.
Yes. Apache 2.0 allows commercial use, modification and redistribution, provided you keep the license and notices and mark your changes.
Not necessarily. The license covers what is released, usually the weights, and the training data and code may still be private.
Reflection AI introduced Beam on Monday, a 501-billion-parameter mixture-of-experts model with 23 billion active, and promised Apache 2.0 weights later in October. Until the weights ship, it is a waitlist and a claim about efficiency.
OpenAI published 722 manuscripts in 372 result families on GitHub on Tuesday, produced mostly by an unreleased internal model, and not all are formally verified. The bottleneck in AI math has moved from finding proofs to checking them.
Nvidia is in talks to acquire open-weight model startup Reflection AI or invest more in it, the Financial Times reported, days after Reflection announced its 501B-parameter Beam model. Nothing changes for you until Beam's weights ship under Apache 2.0.
Reflection AI put Beam's weights on Hugging Face on Oct. 16, a 501B-parameter coding model scoring 81.6% on SWE-bench Verified by its own count. The license is the launch: you can fine-tune it and ship it.
Amazon says TorchNeuron runs existing PyTorch code on its Trainium chips with a single configuration change, released Oct. 20 under Apache 2.0. It attacks Nvidia's switching cost, but Amazon has published no speed or price numbers yet.