Stable Diffusion 3 Released: Open-Source Model Weights Now Available
In this video, Ai Flux discusses the release of Stable Diffusion 3 model weights by Stability AI on June 12th. This release allows users to harness the power of this text-to-image AI for free, particularly for non-commercial use, while commercial use requires a creator’s license. The video highlights that this is a smaller version of the final model, making it suitable for consumer PCs, laptops, and enterprise-tier GPUs. The model comprises two billion parameters and is optimized for photo realism, prompt adherence, and resource efficiency.
The video also mentions the collaboration between Stability AI and AMD, and the availability of a Tensor RT optimized version of Stable Diffusion 3 medium for AMD GPUs. Despite this, Nvidia GPUs are still recommended for running Stable Diffusion due to better performance and tooling like Tensor RT. The weights are available on Hugging Face, and the video shows how to access and use them.
Additionally, the video touches on the model’s strong points, including its ability to handle complex prompts and fine-tuning capabilities. The speaker expresses excitement over the rapid advancements in the AI space, mentioning that the model can also run on Apple M1 processors with reasonable speed. The release of Stable Diffusion 3 is seen as a step towards democratizing access to advanced AI tools, contrasting with more controlled platforms like MidJourney.
Overall, the video provides a comprehensive overview of the Stable Diffusion 3 release, its capabilities, and its implications for the AI and creative communities.