AI TRENDS | Vitalik Buterin Says Local AI Is Nearing Private, High-Task Use on Strix Halo Laptop

Vitalik Buterin said local AI on his Strix Halo laptop is approaching a point where it can handle a large share of tasks while keeping personal context private. According to NS3.AI, a benchmark image showed input-processing rates of 109.82 to 373.22 tokens per second and output rates of 18.42 to 33.37 tokens per second across 10 workloads. The analysis added that wallet software still needs fixed policies, transaction assertions, and human approval before AI can control crypto assets.