AI TRENDS | DeepSeek Plans Mid-July Launch of DeepSeek V4 and Peak-Off-Peak API Pricing

Several developers said they have received an upgrade notification email from DeepSeek. According to Odaily, the email said the official release of DeepSeek V4 is scheduled for mid-July, alongside changes to its API pricing strategy. The email said DeepSeek will introduce peak and off-peak pricing to allocate resources and improve service stability. It added that prices for the V4 Pro and V4 Flash models will remain unchanged during off-peak periods, while API prices during peak hours will be set at twice the off-peak rate. Peak hours were listed as 9:00 to 12:00 and 14:00 to 18:00 each day. For V4 Pro during peak hours, the cached-hit input price was set at 0.05 yuan per million tokens, the cache-miss input price at 6 yuan per million tokens, and the output price at 12 yuan per million tokens. For V4 Flash during peak hours, the cached-hit input price was set at 0.04 yuan per million tokens, the cache-miss input price at 2 yuan per million tokens, and the output price at 4 yuan per million tokens.