Stable Diffusion

Stable Diffusion is a text-to-image diffusion model that Stability AI first released in 2022, built on research from the CompVis group at LMU Munich working with Runway. Unlike closed systems such as DALL-E or Midjourney, its weights were published openly, so anyone with a decent GPU could run it locally, fine-tune it on their own images, or build products on top of it without going through an API. That single decision shaped the entire open generative-image ecosystem that followed, and later versions (SD 1.5, SDXL, SD3) kept extending what the base model could do with resolution, prompt adherence, and image quality.

It's used by hobbyists, artists, researchers, and developers who want direct control over image generation rather than paying per call to a hosted service. Because the model is open, it became the foundation for a large surrounding ecosystem: fine-tuned checkpoints and LoRAs shared on sites like Civitai, ControlNet for guiding pose and composition, inpainting and outpainting workflows, and interfaces like Automatic1111's WebUI and ComfyUI that let people chain these tools together visually.

What sets it apart from most competitors is that openness. Companies can self-host it for privacy or cost reasons, researchers can inspect and modify the architecture, and a community can build training scripts, extensions, and custom models around it in a way that isn't possible with a closed API-only product. It isn't always the sharpest or most prompt-accurate model available, but its flexibility and the size of the tooling built around it keep it central to open-source image generation.