LLM License & Commercial-Use Checker

Quickly check whether an open-weight model allows commercial use. The reference table covers 20+ popular models with their license type, commercial status, and key conditions. Search by model name or license. These are summaries only — always verify the actual license text before shipping a product.

Disclaimer: The information above is a high-level summary for developer reference only and is not legal advice. License terms change, MAU caps may be updated, and your specific use case may trigger conditions not described here. Always read the full license text and consult a lawyer before commercial deployment.

How to use the LLM License & Commercial-Use Checker

Type in the search box to filter the table by model name, family, license type, or commercial status. The table updates live.

  • Commercial: Yes — you can use this model in a commercial product without paying a license fee, subject to the conditions listed in the Notes column.
  • Commercial: Conditional — commercial use is allowed but subject to meaningful conditions, such as monthly active user caps, attribution requirements, or prohibitions on certain use categories.
  • Commercial: No — the model weights are released for research or non-commercial use only. Building a paid product on these weights violates the license.

Click through to the model card or license URL to read the full terms. MAU caps and prohibited-use clauses are the most commonly missed conditions in permissive-looking licenses.

Why open-weight model licensing is complicated

Open-weight models are not the same as open-source software. Most are released under custom licenses that restrict certain uses — even when the weights are freely downloadable. The practical gotchas fall into three categories. First, MAU (monthly active user) caps: Llama\'s community license requires a separate commercial agreement if your application serves more than 700 million monthly active users. Qwen\'s base license has similar carve-outs for large-scale deployments. These caps seem irrelevant at launch but become material fast for successful products. Second, prohibited use categories: Gemma\'s terms, Llama\'s acceptable-use policy, and StarCoder2\'s BigCode OpenRAIL-M all list categories of use that are forbidden regardless of scale — typically weapons, surveillance, and content moderation bypass. Third, derivative work obligations: some licenses require that fine-tuned versions carry the same license, limiting how you can package and distribute customised models.

The distinction between weights and output is also important. Most model licenses govern the weights (the checkpoint files), not the text the model generates. But a few — notably some OpenRAIL variants — attach obligations to the outputs as well. When in doubt, the rule of thumb is: if the license is Apache-2.0 or MIT, you can use it commercially with attribution; if it is a custom license (Llama Community, Gemma Terms, Qwen License, DeepSeek License), read the prohibited-use section carefully before building.

Common use cases

  • Product due diligence — before committing to a model for a commercial product, verify its license status early in the selection process.
  • Switching base models — when swapping a fine-tuned base model, confirm that the replacement carries the same or more permissive commercial rights.
  • Procurement questionnaires — answer enterprise or legal team questions about AI supply-chain licensing with accurate, source-linked summaries.
  • Research vs production split — quickly identify which models are suitable for production (Apache-2.0, MIT) vs research-only (CC-BY-NC, MRL) before allocating infrastructure.
  • Benchmark comparison — when comparing multiple models on quality, simultaneously check license compatibility so you are not optimising toward an unusable option.

Frequently asked questions

Does Apache-2.0 mean fully unrestricted commercial use?

Apache-2.0 is one of the most permissive licenses: you can use, modify, and distribute the model commercially without paying fees. You must retain the Apache-2.0 license notice, provide attribution, and cannot use the licensor's trademarks. There are no MAU caps or prohibited-use clauses beyond patent grants.

What is the Llama Community License?

Meta's custom license for Llama 3.x allows commercial use for products with fewer than 700 million monthly active users. Above that threshold you must request a separate license from Meta. You must also follow Meta's acceptable-use policy, which bans use to build competing AI models and several safety-related categories.

Is Command R or Command R+ usable in a commercial product?

No. Cohere released Command R and R+ weights under CC-BY-NC-4.0 — non-commercial only. For commercial use you need to use the Cohere API or negotiate a separate license with Cohere directly.

What is OpenRAIL-M (BigCode)?

The Responsible AI License - Model variant (RAIL-M) allows commercial use but attaches use restrictions to both the weights and the outputs. Specifically, BigCode OpenRAIL-M prohibits generating code for activities that violate laws, for creating cyberweapons, and for circumventing safety measures. Violations of the attached-use restrictions can terminate your license.

Are fine-tuned models covered by the base model's license?

Generally yes. Most open-weight licenses extend to fine-tunes, merges, and derivative weights. Llama's license explicitly states that derivatives must also carry the Llama Community License. MIT and Apache-2.0 are more permissive about derivative licensing, requiring only that the original license notice be retained.