Ethereum co-founder Vitalik Buterin tested Alibaba's Qwen3.8-Flash-Next on his own laptop without using a cloud server, arguing that local artificial intelligence (AI) is nearing the point where it can handle a large share of everyday tasks. The laptop uses AMD's Strix Halo chip, which combines processor and graphics capabilities with shared memory; configurations offer up to 128 gigabytes, allowing the machine to run a model that carries 125 billion parameters but activates only six billion at a time. Buterin said short prompts produced output at a comfortable reading pace, while very long prompts of tens of thousands of words remained slower. Alibaba published the model's open weights on August 26, and its downloadable version is called Qwen3.8-Flash-Next, although Buterin referred to it as Qwen3.8-Flash. He also cited llama.cpp's improving performance with the model and said the larger Qwen3.8-Max posted strong benchmark results in August. Beyond speed, Buterin proposed using a local model to remove names, wallet addresses or private code before sending a query to a more powerful hosted system. He said this approach could reduce the personal information exposed to cloud providers, though it would not ensure that all sensitive data was removed. The discussion follows his warnings about surveillance during the EU chat control fight and broader calls from crypto users for tighter limits on AI agents. A class action filed in May accuses OpenAI of sharing ChatGPT user queries with Meta and Google. The source concludes that cloud providers still control frontier AI, but improving local performance and increasingly available shared-memory hardware could move more routine work away from their servers, leaving users to decide how much capability to trade for control.