huggingface_hub v1.32.0 ships shared download cache, reusing 16.8GB of weights across repos

vanstriendaniel · x · 2026-09-17

Hugging Face released huggingfacehub v1.32.0 with a shared download cache: downloading Qwen-Image and then Qwen-Image-Edit now reuses 16.8GB of model weights across repos instead of re-downloading them.

Xet verifies file hashes on upload, making cross-repo caching safe. Update via hf update.

Original post →

More from Infra

Infra channel →