According to CoinMeta, Ethereum co-founder Vitalik Buterin stated that he is testing a privacy-enhancing architecture that utilizes personal health and travel data to provide diet and exercise recommendations, while minimizing the risk of revealing private information to remote large models. This solution is coordinated by a local model Qwen 3.8 Flash Next, and it employs a three-layer mechanism to reduce identity exposure: the local model rewrites queries to hide personal information and writing characteristics, zkapi hides payment-related details, and Tor conceals network and IP information. Vitalik claimed that this solution is already operational, but there is still a latency penalty of about 10 to 100 times; however, the speed of the local model is around 20 to 30 TPS. Moreover, the stricter the privacy protection, the less information the remote model can access.