Why we tried BSL
Mergekit began as a research tool for model merging and post-training. Over time, advanced techniques like fusion and tokenizer surgery lived in a private Arcee fork. We wanted to share that work and invest more in the public library, while preventing large competitors from hosting or commercializing it without engagement.
The Business Source License looked like a clean middle path: open enough for the community, protected enough for us. In theory, most developers could continue as before. In practice, any custom license creates gray areas. Even when the answer is "you're fine," the question itself slows adoption. Engineering teams need to route license reviews through legal. Startups hesitate. Contributors wonder if their patches will complicate future use.
That uncertainty did not align with how we want to build.
What changed
During the same period, our scope expanded beyond post-training into the full model lifecycle. We released AFM-4.5B in late July under Apache-2.0. The next generation of models ships in the same permissive terms within weeks. Mergekit now sits inside a much larger training and deployment pipeline. Keeping the library under a custom license while our models use a standard one created misalignment.
Community feedback made the decision straightforward. Developers told us directly: the BSL introduced friction they did not want to navigate. They valued clarity over cleverness. We listened.
What LGPL v3 means in practice
You can use Mergekit freely in commercial or proprietary products. If you modify and distribute the library itself, you must release those changes under LGPL v3. Keep copyright notices intact and provide library source when distributing binaries.
For most teams, this changes nothing operationally. For everyone, it removes the license question from the adoption checklist.
What we learned
Community trust compounds when the license is standard and boring. Custom licenses signal protection. Standard licenses signal confidence. The best way to protect long-term value is to build the best tool and keep improving it.
Clear rules beat clever rules. Simplicity helps teams adopt, integrate, and contribute without routing every decision through legal review. Reducing friction grows the contributor base faster than any protection mechanism.
Alignment across the stack matters. When your models are permissive and your tools are not, you create cognitive overhead for every developer evaluating your ecosystem. Returning Mergekit to LGPL v3 brings the entire pipeline into a single, coherent licensing posture.
What happens next
The repository license file switches to LGPL v3 on Friday, October 31, 2025. Future releases will reflect this change. Existing BSL releases remain in history for transparency, but active development continues under LGPL v3.
If you have been waiting to adopt Mergekit because of licensing uncertainty, that barrier is gone. If you want to contribute, open a pull request or start a discussion in the repository. If you are deploying Mergekit at scale and need guidance, reach out.
To everyone who filed issues, sent patches, and gave us direct feedback about licensing and adoption: thank you. Your candor made this decision easy. We are excited to build the next generation of ML customization tools together, in the open.