Stable Diffusion
The open-source image-generation technology many PixAI models are built on.
Stable Diffusion is the open-source AI image technology that kicked off the modern AI art boom, and it's the foundation many of PixAI's models are built on. You don't need to understand its internals — but the name shows up in model descriptions and Model-Market filters, so here's what it means for you.
The generations you'll encounter
SD 1.5 — the classic generation. Fast and cheap, with a huge library of community models and LoRAs. Works at smaller image sizes.
SDXL — the bigger, better successor: larger images, better anatomy, better prompt understanding. Most modern PixAI models (Haruka, Hoshino, Hinata…) are SDXL-based. See SDXL.
The key practical rule: the generations don't mix. An SD 1.5 LoRA won't work on an SDXL model, and vice versa — see Base-Model.
Beyond Stable Diffusion
PixAI's newest in-house models, Tsubaki and Serin, use a newer architecture called DiT (DiT-Model) rather than Stable Diffusion. All of them share the same core idea though: turning noise into images step by step — see Diffusion-Model.