CNBeta — 2026-05-23#
Top Story#
DeepSeek slashed its V4 Pro model API pricing to just 25% of its original cost, turning a temporary promotional discount into the permanent standard rate. This aggressive move drops the price to 3 RMB per million input tokens for cache misses and 6 RMB for outputs, making it highly competitive for developers integrating AI into programming tools. This ensures developers can freely deploy the model in various applications without cost concerns, further shaking up the competitive landscape for frontier AI models.