All tasks / Models
SUPIR-v0Q / v0F
SDXL-based generative photo restoration and upscaling model from the CVPR 2024 SUPIR paper; non-commercial use only.
- XPixel Group (Fanghua Yu et al.)
- Open weights
Where it ranks
- –Upscaling & restorationModels · Local · no result on this task’s benchmarks yet
Facts
- Vendor
- XPixel Group (Fanghua Yu et al.)
- Family
- SUPIR
- Released
- —
- Weights
- Open: download and run it yourself
- License
- Non-commercial use only (custom declaration)
- GPU memory to run locally
- ~12 GB (README low-VRAM setting: 12 GB for the diffusion model plus 16 GB for the optional LLaVA captioner.)
- Access
- Free (open weights)
- Last verified
- 18 Oct 2026
Price from github.com, 18 Oct 2026.
Benchmark results
No public benchmark results recorded yet.
Where you can use it
No tracked tool lists this model yet; as open weights it runs in local runtimes.