Stable Diffusion 3.5 is now available on Amazon Bedrock. What does it mean for enterprise AI workflows?

While creating AI images that are fun and useful can be a lot of fun, it’s not the only thing enterprises need.

Enterprise Text-to-Image Generation is more than just creating images. It’s all about integrating with existing workflows and enterprise AI tools. Stability AI, the company behind Stable Diffusion understands this direction.

Stability AI and Amazon Web Services announced today that Stable Diffusion is now available on Amazon Bedrock. AWS is the sole public cloud service that offers the flagship StabilityAI models.

This move isn’t about availability alone. It’s all about integration and a new go-to market strategy that will give Stability AI a boost as its new CEO focuses on meeting customer needs. Amazon Bedrock is a single API that allows businesses to access and use a variety of AI models including Stable Diffusion. This is important because AWS’ research shows that many enterprises use multiple models at once. Users like the National Football League and Stride Learning already benefit from this approach. Stability AI is facing a competitive landscape, with competitors such as Google, Midjourney and Black Forest Labs Flux Pro. The company wants to differentiate itself with more image diversity, in terms of style and prompt adherence.

In an exclusive interview with VentureBeat, Prem Akkaraju told VentureBeat that there’s a good reason why Stability AI is on AWS. “It is because that’s where developers and creators are and we want our tools and models to be where they are.” Our goal is empower professional content creators.

Why does AWS need Stable Diffusion when it has its own image-generation tech?

The increasingly crowded text-to-image AI landscape includes models from Amazon. Amazon Nova AI models, including those for image generation, were announced at the beginning of December.

Baskar Sridharan is VP of AI, ML and Infrastructure at AWS. He told VentureBeat the availability of multiple text-to image generation models gives users a choice. Amazon Bedrock offers a single API to users so they can deploy any model available on the platform. Sridharan noted that AWS offers model evaluation tools to help enterprises select the best tool for their deployment.

It’s not surprising that Akkaraju believes Stable Diffusion is superior to other models. Stability AI’s benchmarks on prompt adhesion support this assertion.

Akkaraju said, “We have shown in our research that Stable Diffusion Large leads the market for prompt adherence. This allows models to closely follow the text prompt.

How enterprises can integrate Stable Diffusion with Amazon Bedrock AI workflows

Stable Diffusion version 3.5 large is available to Stability AI customers via the company’s API and Stable Assistant services since late October.

Stability AI does not see a real price difference when using Stable Diffusion via its API or Amazon Bedrock. Stable Diffusion is now available on Amazon Bedrock, which offers enterprise users the benefit of being able to integrate it into larger and more complex enterprise workflows. Enterprises can benefit by a unified workflow which ties together multiple models from different vendors with a single API.

The NFL already uses this approach.

Amazon Bedrock is used by the NFL to create “My Cause, My Cleats”an application that allows users to collaborate and design custom cleats. Sridharan explained the application uses Anthropic Claude as well as Stable Diffusion. The NFL uses Claude for detailed prompts to understand user preferences and determine their wants. This prompt is then fed to Stable Diffusion in order to generate the image. Amazon Bedrock allows you to complete the entire workflow and process without having to switch between services.

Stride Learning, an education vendor, has also benefited from the integration. The company needed images for its online learning game Legends Library – a lot of them, up to 1,000 per minute. Amazon Bedrock offers the infrastructure needed to run Stable Diffusion at this scale. Beyond high-performance scaling, there was a need for secure image output. Amazon Bedrock Guardrails is a great API for this. Sridharan said that Stride Learning can meet AI policies for image creation with guardrails.

Sridharan said that when you use a single API endpoint to do all of this, it is very easy for users to build these types of applications.

Stability AI’s 2024 and its road ahead under the new leadership

Stable Diffusion version 3.5 and its availability on Amazon Bedrock capped off a memorable year for Stability AI.

In March, the company’s founder and former chief executive officer Emad Mostaque announced his resignation due to concerns about lack of focus and revenue. Akkaraju was appointed as a permanent replacement by Stability AI in June.

Akkaraju has been in charge of a series model updates. He also helped bring in new investors, such as Napster founder Sean Parker, and advisors like acclaimed director James Cameron. Akkaraju has a background in visual effects and worked on Cameron’s Avatar. According to him, the professional visual industry will undergo a complete transformation over the next few years, as it moves from rendered content to generated content.

Akkaraju said, “It’s good that we work in the creative industry because they are probably the most demanding clients we could ask for.”

He joked that Stability AI plans to dominate the world. He expects, on a more serious note though, that his company will continue to innovate as it strives to meet the real workflow needs.

Akkaraju said, “We will continue to push this model forward.” “You may even see us release our next image model next year and just stay on top of the wave.”