Stability AI

Stability AI is a London-based open-source generative AI company founded in 2019 by Emad Mostaque. Its 2022 release of Stable Diffusion was the founding shot for open-source text-to-image, breaking image generation out of closed APIs (DALL·E 2 / Midjourney) into a model that anyone could run on a consumer GPU. It catalyzed an entire open ecosystem (Automatic1111 / ComfyUI / SD WebUI).

Key products

Stable Diffusion 1.x (2022): LDM architecture, open text-to-image
Stable Diffusion 2.x (2022-2023): resolution / scale upgrades
SDXL / SDXL Turbo (2023): dual-model pipeline, 1024×1024 high quality
SD3 / SD3.5 (2024): MMDiT architecture, much better text rendering
Stable Video Diffusion (2023): open image-to-video
Stable Audio (2023-): open text-to-music
Stable LM (2023): open LLM, positioned vs Llama
Stable Code (2023-): code LLM
Membership / API platform: commercial cloud

Influence

Single-handedly turned "open-source generative AI" from fringe to mainstream
Catalyzed the Automatic1111 / ComfyUI / SD WebUI ecosystem
Enabled the explosion of ControlNet, LoRA, IP-Adapter fine-tuning techniques
First to make an "open-source generative" business model work (API + subscription + enterprise license)
Emad Mostaque stepped down in 2024-2025; CEO turnover; company shifted to a more conservative commercial path

Limitations

Training-data copyright disputes (lawsuits from Getty Images et al.)
Multiple layoffs and executive turbulence in 2024-2025
Loss of core R&D staff (e.g. Robin Rombach to Black Forest Labs for Flux)
Closed-source flagships (DALL·E 3 / Midjourney v6 / Imagen 3) still lead in quality