Updated October 15, 2026
Stability AI develops generative AI models and services for images, audio, video, and 3D content. Its offerings include the Stable Diffusion image-model family, a developer API, downloadable model weights, and creative applications. Choosing a product means choosing both a capability and a way to run it: a self-hosted model, a hosted API, or an application has different costs and conditions.
The company provides several deployment options. Some models are available under licenses that permit qualifying commercial use without a license fee. Check the exact model and version before building a workflow around it.
Historical Stability AI website screenshot retained from the original 2025 article. It introduces Stable Diffusion 3.5; the current website and model lineup may differ. This is not a GPTOnline-generated or tested result.
What can you create with Stability AI?
- Images: Stable Diffusion 3.5 and image services cover generation and tasks such as inpainting, background removal, outpainting, and upscaling. These can support concept exploration or asset-editing workflows.
- Audio: Stable Audio 3.0 includes Small, Small SFX, and Medium downloadable variants. Large is offered through the API and enterprise self-hosting. Duration and deployment options depend on the variant.
- 3D: Stable Fast 3D and Stable Point Aware 3D work from images to help create 3D assets. Check which models are supported by your chosen service.
- Video: Stable Video Diffusion remains available for self-hosted deployment. The Stable Video API was discontinued on July 24, 2025, so an older video API tutorial may no longer describe an available service.
Choose the right access route
Self-hosting suits developers who want to run and customize an available model in their own environment. Budget for suitable hardware, installation, storage, maintenance, and the applicable model license.
The API suits applications that need hosted generation or editing. It requires an account, an API key, and service credits. You manage your integration and usage budget rather than the model-serving infrastructure.
Applications provide a ready-made working interface. Brand Studio focuses on brand-oriented creative production, with features and access varying by plan. Stable Audio offers a web app and a macOS AU/VST3 instrument plugin for supported music-production software. Check current compatibility before installing; the announcement describes a beta experience.
Pricing and licensing
Self-hosted models
The current licensing page presents a free Community license for qualifying use and a custom-priced Enterprise option. The old Professional membership price of US$20 per month should no longer be used as a general current price for Stability AI.
Under the Community License Agreement, commercial users must register. Its revenue trigger counts the user and affiliates, individually or together, and includes revenue unrelated to Stability AI. Commercial users exceeding US$1 million in annual revenue, or its local-currency equivalent, must obtain a separate license. Confirm eligibility directly when approaching the threshold.
Coverage is limited to the listed Core Models; other releases can have different licenses. Attribution, redistribution, acceptable-use, and other conditions still apply. A free license does not cover your computing costs or automatically include hosted services.
Hosted API usage
The developer documentation prices purchased credits at US$1 per 100 credits, equivalent to US$0.01 each. Representative published API rates are:
| Image service | Credits per successful generation | Equivalent USD cost |
|---|---|---|
| Stable Image Core | 3 | US$0.03 |
| Stable Image Ultra | 8 | US$0.08 |
| Stable Diffusion 3.5 Large | 6.5 | US$0.065 |
These are usage prices, not monthly allowances. Editing, upscaling, audio, and 3D services have their own rates. Check the current price page and endpoint documentation before estimating a production budget.
Creative applications
Application subscriptions have their own credit allocations and conditions. Brand Studio’s official plan overview describes trial, Creator, Core, and Enterprise options. Its former marketing price page is unavailable, so verify the offer in the application before purchasing. Do not apply the API’s dollar-per-credit conversion to an app subscription.
Example workflow: a fictional product campaign
Imagine a small design studio preparing a launch concept for a fictional reusable water bottle. This is an illustrative brief, not a product test.
- Define the deliverable: three square concept images showing the bottle in a bright kitchen, a travel bag, and an outdoor setting.
- Choose the route: use an application for a designer-led session, or the API if the studio is building a repeatable workflow.
- Set a starting budget: 30 successful Stable Image Core generations would consume 90 credits, equivalent to US$0.90 at the listed rate. Additional edits or upscales increase the cost.
- Review before delivery: check bottle shape, label spelling, reflections, composition, and consistency across images. Use approved brand assets and confirm usage rights before publication.
Important limitations
Model availability, application plans, and API prices can change independently. Downloadable weights do not make every release unrestricted open-source. Output rights remain subject to applicable law and terms; a model license does not guarantee that every generated asset is suitable for commercial publication.
This overview is based on public documentation, without hands-on generation testing. Evaluate a small representative brief before committing to a larger workflow. For other options by task, browse our AI tools directory.
