It’s neat how when OpenAI or another model provider creates a new model (e.g. text-to-video like Sora), the open source community gets to work and produces an open source version. This happened with Sora (although the quality is not quite there yet):