Stable Diffusion Models Family: SD 1.5, SDXL, SD 3.5 for ComfyUI
Stable Diffusion lineage from Stability AI: SD 1.5 (Runway), SD 2.x, SDXL, and SD 3.5. Open text-to-image checkpoints widely used in ComfyUI.
S
Stable Diffusion by Stability AI
Text-to-ImageOpen SourceControlNetOpen text-to-image model lineage from Stability AI. Covers classic UNet checkpoints (1.5 / 2.x / SDXL) through MMDiT-based SD 3.5. SD 1.5 weights follow the Stability / Runway release lineage still common with community fine-tunes and ControlNet v1.1.
Versions
| Version | Architecture | Native res | Highlights |
|---|---|---|---|
| SD 1.5 | Latent Diffusion (UNet) | 512×512 | Stability / Runway lineage; ControlNet v1.1 ecosystem |
| SD 2.0 / 2.1 | Latent Diffusion (UNet) | 768×768 (2.1) | Official 2.1 checkpoints; dedicated SD 2.1 ControlNets (thibaud) |
| SDXL | Latent Diffusion (UNet) | 1024×1024 | Base + Turbo + Refiner; community SDXL ControlNet collection |
| SDXL Turbo | Latent Diffusion (UNet) | 512×1024 | 1–4 step distilled generation via SDTurboScheduler |
| SD 3.5 | MMDiT | 1024×1024 | Medium / Large / Turbo + Comfy-Org FP8 packs |
| Stable Cascade | 3-stage (DiT+DiT+VAE) | 1024×1024 | Stage C → Stage B → Stage A; ControlNet support |
| Stable Video Diffusion | Latent Diffusion (UNet) | — | Image-to-video: SVD (14f) / SVD-XT (25f) |
Select a version for tutorials and related news. Curated weight downloads live in the Model Hub catalog.
SD 1.5 lineage: The widely used
stable-diffusion-v1-5checkpoints descend from Stability AI training with Runway collaboration on the v1 line. They remain the default base for many ControlNet v1.1 and community fine-tune workflows in ComfyUI.