Format Notice: Except for SAM3, all models in this repository are provided in INT8 format. Some of these INT8 versions were converted by me using the following tools:

Note on my quantized/ INT8 conversions:

๐Ÿชถ = Custom quantized / Int8 ConvRot converted version,

Based on testing, INT8 quantized versions of SD1.5 and SDXL architectures (including Pony, Illustrious, etc.) are actually slower than the original models due to VRAM bandwidth bottlenecks and the overhead of runtime dequantization. Since SD1.5 is already lightweight and SDXL architectures are already fast, the speed gains from quantization are negligible.In my simple benchmarks, only the Illustrious (IL) INT8 version showed a positive speed optimization, while all other INT8 versions were slower.

Furthermore, for SDXL variants, the dual-CLIPs are tightly integrated with the UNetโ€”splitting them would cause severe storage bloat (e.g., requiring 60 CLIP-L+ 60 CLIP_G files for 60 checkpoints). However, I still provide these converted versions specifically for low-VRAM environments or testing, allowing users with limited hardware to run models that would otherwise OOM."

Qwen Image Edit (Click to collapse)
SAM3 (Click to collapse)
  • ๐Ÿชถsam3_*.pt: FP16/Mixed FP8/FP8 version, compared with the official FP32 version.
    • Source: SAM3
    • Location: ๐Ÿ‘‰ Open Folder
    • Recommended Nodes: ComfyUI-Easy-Sam3 or ComfyUI-RMBG
    • File Placement:
      ๐Ÿ“‚ ComfyUI/
      โ”œโ”€โ”€ ๐Ÿ“‚ models/
      โ”‚   โ”œโ”€โ”€ ๐Ÿ“‚ sam3/
      โ”‚   โ”‚   โ”œโ”€โ”€ ๐Ÿ“ฆ sam3_*.pt
      
    • Note: RMBG only reads files named exactly as sam3.pt, while ComfyUI-Easy-Sam3 can read any model in the ComfyUI/models/sam3 folder.
Moss Image (Click to collapse)
Krea 2 (Click to collapse)
FLUX.2-klein-9B (Click to expand)
Anima (Click to expand)
Flux.1-dev (Click to expand)
Illustrious (Click to expand)
Pony Diffusion V6 (Click to expand)
SDXL (Click to expand)
Stable Diffusion 1.5 (Click to expand)
Downloads last month
799
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Space using waveshaker/comfy_models 1