Small startup shifts programmers’ schedules to cut AI token costs

A startup with about a dozen employees has begun scheduling programmers around cheaper AI usage periods rather than maintaining round-the-clock server coverage. The company uses coding services from DeepSeek, Zhipu and other providers, including MiniMax, GLM and Volcano Engine. Employees must take one weekday and one weekend day off each week, with the days chosen freely but scheduled in advance, while lunch breaks have been pushed to after 2 p.m. DeepSeek’s latest API pricing sets weekday peak rates at twice its off-peak rates and changed weekends to all-day off-peak pricing on Aug. 23. Zhipu’s Coding Plan also varies usage allowances by time, with model calls during off-peak periods deducted at 50% of the standard credit rate. Some commenters said the practice showed people adapting their work schedules to AI pricing, while another said their employer was considering a similar approach.

本网站上的信息是使用AI生成的,我们无法保证其准确性。 请仅作为参考信息使用。