Stable Diffusion API for Developers

Production Stable Diffusion, One REST Endpoint

The ModelsLab Stable Diffusion API gives developers programmatic access to the full Stable Diffusion family — SDXL, SD 1.5 and SD 3 Medium — plus 10,000+ community checkpoints and LoRAs, all through a single REST endpoint. You pass a model_id and a prompt to /api/v7/images/text-to-image and receive hosted image URLs in JSON within a few seconds.

Stable Diffusion is the most widely used open image-generation model family, and running it well in production means managing GPUs, model loading, VRAM, and scaling. ModelsLab handles all of that: popular models stay warm, a rarely used checkpoint or LoRA loads onto a GPU on first use (a few minutes), and auto-scaling absorbs traffic spikes. You keep the Stable Diffusion workflow you already know and drop the infrastructure.

Beyond base models, the API exposes the tools Stable Diffusion developers rely on — ControlNet for structural guidance, LoRA for style and subject fine-tuning, and dedicated inpainting and image-to-image endpoints. The same API key also covers FLUX and 10,000+ other image, video, audio, and LLM models when your product grows.

What You Can Build with the Stable Diffusion API

The Stable Diffusion API supports the full range of image workflows through one key:

• Text-to-Image — Generate images from prompts with SDXL, SD 1.5 or SD 3 Medium at up to 2048×2048
• Image-to-Image — Transform existing images with strength-controlled, prompt-guided edits
• Inpainting — Mask a region and regenerate it from a prompt for object removal and replacement
• ControlNet — Guide generation with pose, depth, edge, and segmentation reference maps
• LoRA & Custom Checkpoints — Attach community LoRAs or upload and serve your own fine-tuned models
• Upscaling — Enhance output resolution with super-resolution models through the same API

Stable Diffusion API Pricing

The Stable Diffusion API uses simple pricing: $0.0047 per image from the usage included in your plan ($47 on Standard covers 10,000 images), the same across SDXL, SD 1.5 and SD 3 Medium with no resolution upcharge up to 2048×2048. Stability AI’s own API lists SDXL 1.0 from $0.009 per image; Replicate bills SDXL by GPU time at about $0.0026 per image, which is cheaper than any ModelsLab plan below about 57,000 images a month (prices checked 2026-09-27).

For high-volume workloads, the $149/month Open Source Unlimited plan removes per-image cost entirely — the cheapest option at scale. Paid plans start at $21/month (Basic) and $47/month (Standard), and dedicated enterprise endpoints with custom SLAs start at $249/month. Compared with self-hosting, you avoid GPU rental, VRAM tuning, and on-call ops entirely.

API Response Format

The Stable Diffusion API returns a consistent JSON response across all endpoints. Generated images come back as publicly accessible URLs (available for 24 hours) or as base64 data for direct embedding. Synchronous requests return images in the response; for longer or batched jobs, use asynchronous mode and poll the fetch endpoint with the returned request_id, or register a webhook to receive completions.

Integration and SDKs

ModelsLab provides SDKs and integration paths for the Stable Diffusion API across popular stacks:

• Python SDK — pip install modelslab — full endpoint coverage with async support
• JavaScript/TypeScript — npm install modelslab — works in Node.js, Deno, and the browser
• REST API — standard HTTP endpoints usable from any language (cURL, Go, Ruby, PHP, Java)
• Webhook callbacks — receive generation completion notifications for async pipelines
• Documentation — per-parameter guides and examples at docs.modelslab.com