Client-side Gen AI: Performance demo, no worker ⏱️

This demo showcases client-side Gen AI (in-browser) without a worker. This web page downloads the Gemma 2B model, and uses it through the MediaPipe LLM Inference API.

The animation below won't keep running smoothly throughout your LLM usage. Both the model preparation steps and the inference work take place on the main thread. That's a performance bottleneck.