Stable Diffusion 3.5 is the latest version of Stability AI's foundational open-source image model. With the largest community of LoRAs, ControlNets, and custom checkpoints, it offers unmatched customization and control over AI image generation.
Most Customizable AI Image Generator — SD 3.5 Large/Medium by Stability AI.
Access hundreds of thousands of custom checkpoints, LoRAs, and styles on CivitAI and Hugging Face.
Precise control over pose, depth, edges, and composition through ControlNet conditioning.
All weights publicly available with permissive licensing for personal and commercial use.
Edit specific image regions or extend images beyond their borders with seamless results.
Transform existing images using text prompts while preserving composition and key elements.
Train your own LoRAs on specific styles, brands, or characters in just hours on consumer hardware.
Try via DreamStudio (web), or install ComfyUI / Automatic1111 / Forge for local use with full control.
Browse CivitAI for thousands of specialized models. Choose one matching your desired style or subject.
Write detailed prompts with style tokens. Use negative prompts and ControlNet for precise results.
Create game characters, environments, and items with consistent style using custom LoRAs.
Use inpainting and outpainting for advanced photo manipulation impossible with cloud-only tools.
Train brand-specific models that consistently match your visual identity across all generated assets.
Unlimited self-host. Customization optimized.