[AI Begins to Impact Work Schedules: A Company Implements Shift Work for Programmers to Save Tokens and Reduce Peak Usage Costs] Beating AI News Flash: A startup with around ten employees has started implementing shift work for its programmers—not because servers need 24/7 monitoring, but to save on AI tokens. An employee posted on V2EX, stating that the company heavily uses coding services like DeepSeek and Zhipu, where model usage during peak hours incurs higher fees or token consumption. To cut costs, the company decided to schedule work during off-peak hours. The new attendance policy allows employees to take one weekday off and one weekend day off each week, with flexible combinations, but schedules must be arranged in advance. Additionally, weekday lunch breaks have been postponed to after 2:00 PM. The poster later revealed that the company, with just over ten employees, currently subscribes to multiple coding plans, including MiniMax, GLM, DeepSeek, and Volcano Engine. DeepSeek's latest API pricing sets peak weekday rates at twice the off-peak rates, and as of August 23, weekends have been changed to all-day off-peak pricing. Zhipu's Coding Plan also calculates token consumption based on time, with model calls during off-peak hours consuming only 50% of the base credits. In the comments under the original post, some people lamented that humans are now actively adjusting their work schedules to align with AI pricing. Others mentioned that their own bosses have heard of similar practices and are considering whether to follow suit. [Original Link]