Stability API

Converts text prompts into images using Stable Diffusion models with fine-grained control over generation parameters including sampling steps, guidance scale, seed, and model selection. The API accepts text descriptions and returns generated images in PNG or JPEG format, with support for negative prompts to exclude unwanted elements. Generation is performed server-side on GPU infrastructure with configurable inference parameters affecting quality, speed, and determinism.

Transforms an existing image based on a text prompt while preserving structural elements and composition. The API accepts an input image and text prompt, applies diffusion-based editing with a configurable strength parameter (0-1) controlling how much the original image influences the output, and returns a modified image. This enables style transfer, content modification, and guided image evolution while maintaining spatial relationships.

Returns structured error responses with specific error codes, messages, and diagnostic information for failed requests. The API distinguishes between client errors (invalid parameters, authentication failures), rate limiting, and server errors, providing actionable feedback for debugging. Error responses include error codes, human-readable messages, and sometimes suggestions for remediation (e.g., 'reduce steps' for timeout errors).

Provides specialized model variants trained on specific visual domains (photography, illustration, 3D rendering, anime, etc.) that can be selected to influence generation style without explicit style prompting. The API routes requests to domain-specific models based on selection, enabling consistent aesthetic output aligned with training data characteristics.

Exposes generation capabilities through RESTful HTTP endpoints with standardized JSON request/response payloads, authentication via API keys, and consistent error handling. The implementation follows REST conventions with POST endpoints for generation requests, GET endpoints for status/results, and structured error responses with detailed error codes and messages.

Generates new content within masked regions of an image while preserving unmasked areas. The API accepts an image, a binary mask (or alpha channel), and a text prompt, then applies diffusion-based inpainting to fill masked regions with content matching the prompt. The mask defines which pixels can be modified (white) vs preserved (black), enabling targeted content replacement, object removal, or insertion without affecting surrounding areas.

Extends images beyond their original boundaries by generating new content that matches the style and context of the existing image. The API accepts an image and optional prompt, then expands the canvas in specified directions (up, down, left, right) with AI-generated content that maintains visual coherence. This enables expanding compositions, adding background context, or creating panoramic variations without manual editing.

Increases image resolution while enhancing details and reducing artifacts using AI-based upscaling. The API accepts an image and target upscaling factor (2x, 4x, etc.), applies a specialized upscaling model that reconstructs high-frequency details, and returns a higher-resolution version. The upscaling process uses diffusion or super-resolution techniques to add plausible details rather than simple interpolation, improving perceived quality.

Generates short video clips from text descriptions using diffusion-based video synthesis models. The API accepts a text prompt and optional parameters (duration, resolution, frame rate), then generates a coherent video sequence where frames are synthesized to match the prompt while maintaining temporal consistency. The model ensures smooth motion and coherent object tracking across frames rather than generating independent frames.

Provides access to multiple Stable Diffusion model variants (e.g., SDXL, SD 1.5, SD 3) with different performance characteristics and quality profiles. The API allows specifying which model to use per request, enabling developers to choose between faster inference (smaller models) and higher quality output (larger models). Each model has different parameter ranges, supported features, and latency profiles, requiring explicit selection based on use case requirements.

Supports asynchronous batch image generation through job submission and polling APIs. Developers submit generation requests with a callback URL or polling endpoint, receive a job ID, and retrieve results when processing completes. This enables high-throughput image generation without blocking on individual request latency, suitable for processing large image queues or integrating with background job systems.

Exposes detailed generation parameters with model-specific valid ranges and defaults, including guidance scale (controlling prompt adherence), sampling steps (affecting quality vs speed), seed (for reproducibility), and sampler selection (different diffusion sampling algorithms). The API validates parameters against model-specific constraints and returns errors for out-of-range values, requiring developers to understand parameter semantics and model capabilities.

Provides image generation capabilities through standard REST API endpoints accepting JSON payloads and returning image data or JSON responses. The API uses HTTP POST for generation requests, supports standard HTTP status codes and error responses, and integrates with any HTTP client library or framework. Authentication uses API keys passed in request headers, following standard REST conventions for stateless request/response cycles.