DeepSeek Introduces Peak-Hour API Pricing for LLM Inference

DeepSeek's new API pricing, effective 16:00 UTC on August 16, 2026, splits LLM inference costs into peak and off-peak windows — a time-of-use model long standard in electricity grids and cloud spot markets but new to AI API billing. Peak hours run 01:00–04:00 and 06:00–10:00 UTC; off-peak rates are 50% lower. Overall increases range from 50% to 1,100% above current prices. Whether OpenAI, Anthropic, or Google adopt similar surcharging remains an open question.