SAMA 14b - Video Editing Model based off Wan 2.1 (Apache 2.0)

Developer Cynthiazxy123 has launched SAMA 14B, a significant open-source video editing model built on the Wan 2.1 architecture. Released under the permissive Apache 2.0 license, this 14-billion parameter model represents a major step toward democratizing AI-powered video editing. Available immediately on both GitHub and Hugging Face, SAMA 14B allows developers and creators to access sophisticated video manipulation capabilities that were previously locked behind proprietary platforms and expensive APIs.

The model enables direct text-to-video editing, where users can describe changes they want to make—like "remove the car from this scene" or "apply a cyberpunk style"—and the AI executes these edits on the video content. This approach bypasses the need for complex manual editing software, making professional-grade video manipulation accessible to a broader audience. The open-source nature means developers can fine-tune the model for specific use cases, integrate it into custom workflows, or build commercial applications without licensing restrictions.

SAMA 14B's release comes at a time when video content creation is exploding across social media and professional platforms. By providing an open alternative to closed systems like Runway ML or proprietary features in Adobe's tools, this model could accelerate innovation in video editing workflows. Early adopters are already experimenting with applications ranging from content creation and marketing to educational video production and automated video preprocessing for larger AI pipelines.