Stability AI Unveils Stable Diffusion 3 Medium: A Powerful, Efficient Text-to-Image AI Model

Stability AI has released a smaller version of its Stable Diffusion 3 (SD3) artificial intelligence (AI) model, dubbed Stable Diffusion 3 Medium. This new text-to-image model retains the functionality of the larger SD3 model but with lower GPU requirements and reduced power consumption, making it more accessible for use on consumer-grade PCs and laptops.

Key Takeaways

• Stable Diffusion 3 Medium is Stability AI’s most advanced text-to-image open model yet.

• The small size of this model makes it perfect for running on consumer PCs and laptops as well as enterprise-tier GPUs. It is suitably sized to become the next standard in text-to-image models.

• The weights are now available under an open non-commercial license and a low-cost Creator License.

• To try Stable Diffusion 3 models, try using the API on the Stability Platform, sign up for a free three-day trial on Stable Assistant, and try Stable Artisan via Discord.

The Stable Diffusion 3 Medium model has two billion parameters, compared to the eight billion parameters in the larger SD3 model. Despite the smaller size, Stability AI claims that the new model will deliver similar levels of efficiency and performance, including detailed photorealistic outputs, high-quality flexible styles, and improved realism in areas like hands and faces.

The minimum requirement for running the Stable Diffusion 3 Medium model is 5GB of GPU VRAM, with a recommended 16GB of VRAM. This is a significant improvement over the higher GPU requirements of the larger SD3 model, making it more accessible to a wider range of users.

In addition to the improved efficiency, the Stable Diffusion 3 Medium model also retains the advanced capabilities of its predecessor, including the ability to understand complex prompts with spatial reasoning, compositional elements, actions, and styles. The company has also improved the model's handling of typography, which has been a common challenge for image generation models.

Stable Diffusion 3 Medium is now generally available through Stability AI's Fireworks AI-powered API, the Stable Assistant platform, and the Stable Artisan Discord server. The open weights for the model have also been made available on Hugging Face with a non-commercial license, while commercial use requires a creator license from the company.