Nvidia Launches Cosmos 3 Super Models That Generate Images and Video 25 Times Faster
Nvidia released a new version of its Cosmos 3 Super generative models that use a 4-step process to produce images and video up to 25 times faster than the original releases. The company stated the updated models rank first in an open-weight benchmark for image-to-video generation and second in the text-to-image category.
The new models are now available on the Hugging Face platform. Nvidia’s announcement highlights a focus on reducing the computational steps required for generative media, making faster video and image generation accessible to developers.
From the sources (2 posts)
@nvidiaaiThe new 4-step Cosmos 3 Super models generate images and video up to 25x faster than the originals, and still rank among the best open-weight models on @ArtificialAnlys. 🥇 #1 for image-to-video (no audio) 🥈 #2 for text-to-image Try them o
@victormustarRT @NVIDIAAI: The new 4-step Cosmos 3 Super models generate images and video up to 25x faster than the originals, and still rank among the…