Google’s Pixel-side AI speedup avoids retraining the deployed model. By adding a frozen Multi-Token Prediction path to Gemini Nano v3 on Pixel 9 and 10, Google reports 50% or greater token-generation speedups and 130MB less memory than a standalone drafter.
The thread split between the convenience of “local LLM in Chrome” and corrections about WebGPU acceleration, model identity, and browser-controlled limits.
Google Chrome has been quietly installing a 4GB Gemini Nano AI model on user devices without consent. The file reinstalls itself after deletion, raising GDPR violation concerns and significant environmental impact questions.
HN saw the appeal immediately: local prompts, no API keys, more privacy. The thread turned just as quickly to the friction points, especially the storage and hardware bill attached to browser-side AI.