Checking WebGPU and shader-f16…
First run: about 1.1 GB of model files from Hugging Face. Images and answers stay in this browser. Model answers can be wrong.
Your answer will appear here.
Runs locally using FastVLM-0.5B ONNX, Transformers.js 4.3.0 and WebGPU. For another runtime, see the Python guide.