Repository navigation
Supported Model Types
eduardoabreu81 edited this page Aug 19, 2026
·
1 revision
| Architecture | Notes |
|---|---|
| SD 1.x / SD 2.x | Classic Stable Diffusion |
| SDXL | Base SDXL and derivatives |
| Pony | Pony V6 and variants |
| Illustrious | Illustrious XL |
| NoobAI | NoobAI (Illustrious-based) |
| FLUX | Dev, Krea, Kontext, Klein |
| Krea 2 | Krea 2 turbo and related variants |
| Wan | Wan 2.2 — text/image to video |
| Qwen | Qwen-Image, Qwen-Image-Edit |
| Z-Image | Z-Image, Z-Image Turbo |
| LTXV | LTXV video/image model family |
| Lumina | Lumina-Image 2.0 |
| Anima | Anima |
| Chroma | Chroma1-HD |
| Ernie | Ernie-Image / Ernie-Image-Turbo |
| Cascade | Stable Cascade |
| SVD | Stable Video Diffusion |
| Hunyuan | Hunyuan |
| Other | Catch-all; fully configurable |
File formats: .safetensors, .ckpt, .gguf, .pt, .pth, .bin, .th, .vae, and .zip (Wildcards).
Krea 2 note: the
krea2_turbo_fp8_scaled.safetensorscheckpoint requires the Qwen-Image VAE (qwen_image_vae.safetensors) and the Qwen3VL 4B text encoder (qwen3vl_4b_fp8_scaled.safetensors) rather than FLUX text encoders. Place both files in the Forge Neomodels/text_encoder/folder.
Custom categories can be defined in Settings → Model Organization using a simple JSON pattern list.
Related: Auto-Organization · Browser Sources