StabilityAI has introduced Stable Diffusion 3.5, enhancing its AI image-generation model with improved realism, accuracy, and stylization. The model is available for free for non-commercial purposes, such as academic research and for small and medium businesses with annual revenues below $1 million, according to Tom’s Guide.
Configurations and Customization Options
Stable Diffusion 3.5 is offered in three configurations: Large (8B), Large Turbo (8B), and Medium (2.6B). These models are optimized for consumer hardware, ensuring accessibility for a wide range of users. All configurations are customizable, supporting different creative needs.
In its press release, StabilityAI addressed the feedback on the earlier Stable Diffusion 3 Medium model, which fell short of community expectations upon its June release. “After listening to valuable feedback, rather than a quick fix, we decided to take the time to develop a version that advances our mission to transform visual media,” the company explained.
Features, Performance, and Licensing
The new version emphasizes high performance, flexibility, and a variety of stylistic results. Users can apply settings like photography or painting modes and utilize hashtags for specific styles, such as boho, impressionism, or modern. Keywords in queries also help generate more realistic images.
Among the models, the Large configuration leads in query matching and image quality, adds NIX Solutions. The Turbo version offers the fastest rendering time, while the Medium configuration balances quality and speed, making it ideal for content creation. StabilityAI confirmed that all configurations are available under the Stability AI Community License, with commercial use requiring an Enterprise License.
We’ll keep you updated as StabilityAI continues to enhance its offerings.