Introducing @huggingface/kernels: 200+ WebGPU Kernels for Local AI
On September 1, 2026, Hugging Face released @huggingface/kernels, a JavaScript library for loading and running optimized WebGPU kernels from the Hub, along with an initial collection of 207 kernels and a new in-browser benchmarking tool called Fleet.
What happened
Event details
Hugging Face announced the release of @huggingface/kernels, a minimal JavaScript library for loading and running optimized WebGPU kernels from the Hugging Face Hub. The release includes an initial collection of 207 kernels at huggingface.co/webgpu-kernels, covering operations used across a wide variety of machine learning architectures and workloads. Each kernel is published as a complete, versioned package, with its interface, shader templates, correctness cases, benchmark cases, and usage instructions all living together on the Hub. The kernels are Apache-2.0 licensed and published as individual repositories in the webgpu-kernels organization. Hugging Face also launched Fleet, an in-browser GPU benchmarking and testing suite that runs and scores kernels on user hardware. With user consent, every run adds private evidence that can help find failures, improve kernel variants, and make better optimization decisions across real-world hardware. Running these kernels requires a browser with WebGPU support; WebGPU availability depends on the browser, operating system, GPU, and driver. Developers can install the package from npm with @huggingface/kernels@preview and invoke kernels using getKernel.
Assessment
Why it matters
This is an official Hugging Face release of a foundational WebGPU kernel library and a 207-kernel collection, plus a new benchmarking tool (Fleet), representing foundational infrastructure for browser-based local AI inference.
Availability
Access notes
The library is available via npm as @huggingface/kernels@preview; the kernel collection is hosted at huggingface.co/webgpu-kernels. Running kernels requires a browser with WebGPU support.