David· @DavidSHolz · X·· 5 小时前AI 评分20
AI 导读
假设一个前沿模型是跑在 GB300 NVL72 上的 10T MoE。那大约是 ~5 焦耳/token，或者按人类思考速度算约 ~50 W，而我们大脑是 20 W。电网电价是 $0.08/kWh，吃米饭是 $0.50/kWh。所以从每次思考的能耗角度来看，我们现在和前沿模型惊人地相似！
正文
Let's say a frontier model is a 10T MoE on GB300 NVL72. That's ~5 Joules/token or ~50 W at human thinking speed vs our brain at 20 W. The energy grid is $0.08/kWh and eating rice $0.50/kWh. So on an energy-per-thought standpoint rn we're surprisingly similar to frontier models!
来源：David · x.com