Hugging Face Blog·· 2026-07-23
Bringing Nunchaku 4-bit Diffusion Inference to Diffusers
Bringing Nunchaku 4-bit Diffusion Inference to Diffusers
AI summary
Hugging Face introduced Nunchaku Lite into Diffusers, allowing Nunchaku quantised checkpoints to load directly with from_pretrained(), without custom pipelines or local CUDA compilation.
Selection record
Threshold 60Official, first-handFirst 58Second 62
AdmittedSum of both 120 ≥ twice the threshold 120
- Source tier
- Official, first-hand; this tier's threshold is 60
- Pre-filter
- passed：Diffusers集成Nunchaku 4-bit扩散推理量化技术
- Why it was chosen
- Native 4-bit support clarifies memory and latency trade-offs for running diffusion models on consumer GPUs.
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Hugging Face Blog · huggingface.co