Commercial / SaaS licensing clarification
Hi dots.mocr team,
First, great work on the model — the benchmark results are impressive and the multilingual capability is genuinely best-in-class at this size.
I'm planning to build a commercial SaaS product for PDF extraction and RAG pipelines, and dots.mocr is my top choice for accuracy. Before proceeding, I'd like to clarify a few licensing questions:
1. Dual license conflict
The repo contains two license files — LICENSE (MIT) and dots.mocr LICENSE AGREEMENT (custom). These appear to have contradictory terms in some areas. Which one governs the model weights, and which governs the code? How should conflicts between them be resolved?
2. Commercial / SaaS use of model weights
Is it permitted to use the dots.mocr model weights in a closed-source, commercial SaaS product where end users access the model's output via an API? If yes, are there any conditions (attribution, revenue thresholds, etc.)?
3. PyMuPDF / AGPL dependency
The requirements.txt includes PyMuPDF, which is AGPL-licensed. This is incompatible with closed-source commercial products. Is there a plan to make this an optional dependency, or can you confirm which parts of the pipeline require it so we can swap it out?
Happy to reach out via any other channel if preferred. Thank you!