Stability AI Releases Stable Diffusion 3.5: Open Source Image Generation Leaps Forward

Stability AI released Stable Diffusion 3.5, its most capable open-source image generation model. SD 3.5 Large (8.1B parameters) achieves quality competitive with Midjourney V6 and DALL-E 3 while remaining fully open-source under a research/commercial license. Key improvements: significantly better text rendering (previously SD's weakness), improved human anatomy, better instruction following, and new 2MP resolution support. Available on Hugging Face and via API at $0.065/image. The open-source release enables self-hosted deployment without per-image API costs—important for privacy-sensitive and high-volume use cases.