"With Nebius, we’re able to efficiently utilize clusters of L40S GPUs for NOVA-1's video inference for businesses. It is incredibly efficient — we see 40% cost efficiency gains with L40S without sacrificing content quality or video generation speed. Our consumer-targeted model was initially trained on Nebius infrastructure, and now hundreds of thousands of users are generating personalized videos on the Diffuse app, which is pioneering​ AI-powered social media content creation on mobile devices."