Stable Diffusion 3 API Launch: Get Ready for the Upcoming Stable Assistant Release!

After just two months in preview, Stability AI is excited to launch its next-generation Stable Diffusion 3 generative AI model, alongside an early preview of its innovative chatbot technology, dubbed “Stable Assistant.”

Initially announced in February as a preview, Stable Diffusion 3 is now accessible via an API on the Stability AI developer platform. This API enables easy integration of the model's powerful text-to-image generation capabilities into various services and applications. Additionally, a turbo variant—Stable Diffusion 3 Turbo—has been introduced for enhanced performance.

With Stable Diffusion 3, Stability AI employs advanced machine learning techniques aimed at significantly improving image and typography quality. A primary focus during the API release has been to ensure the model is production-ready.

“We have implemented numerous safeguards to prevent misuse of SD3, continuously refining these measures based on user feedback,” said Christian Laforte, CTO and interim co-CEO of Stability AI.

Open Model Coming Soon

While Stable Diffusion 3 is now available via API, an open model is not yet released but is on the way. “We will continuously improve the model before its open release,” Laforte affirmed. “In line with our commitment to open generative AI, we will soon make the model weights available for self-hosting through a Stability AI Membership.”

This membership strategy, first announced in December, aims to establish a new revenue model for the company.

Fireworks Partnership Enhances API Performance

Stability AI's partnership with Fireworks AI will enhance the performance of the Stable Diffusion 3 API. Optimizing API inferencing for generative AI applications—especially at scale—can be complex, but Fireworks AI’s expertise in machine learning compilers will help address these challenges.

“Fireworks AI are industry-leading ML compiler experts, a vital component for optimizing our models’ inference speed,” Laforte noted. “Partnering with them allows us to deliver the fastest and most reliable enterprise-grade API platform in the market.”

Innovations in Stable Diffusion 3

At the core of Stable Diffusion is the diffusion model, with several innovations enhancing its capabilities. Notably, the introduction of the Multimodal Diffusion Transformer (MMDiT) architecture improves text understanding and typography accuracy.

For the SD3-Turbo model, which offers faster performance, a novel method called Latent Adversarial Diffusion Distillation (LADD) is employed. “Essentially, SD3-Turbo is up to 10 times faster than SD3 while producing images that are nearly as high-quality,” Laforte explained.

Introducing Stable Assistant

In addition to the new Stable Diffusion model, Stability AI has unveiled an early beta of Stable Assistant, a chatbot powered by the company’s text and image generation technology. Similar to OpenAI’s ChatGPT Plus’s integration with DALL-E 3, Stable Assistant enables image generation through conversation.

Laforte describes Stable Assistant as a user-friendly chatbot that combines the capabilities of Stable Diffusion 3 and Stable LM 2 12B, which was recently released. This tool not only generates images from discussions but also provides informative responses, aids in writing projects, and enhances content with relevant images.

“Stable Assistant aims to become our multimodal chatbot, offering access to all our models and API services without requiring technical expertise,” Laforte stated. “We plan to keep enhancing its capabilities by adding image editing and incorporating models from other modalities, including video, 3D, audio, and code.”