DeepSeek Makes 75% Price Cut on V4 Pro Permanent — $0.003625/M Input Tokens, Deepening Global AI Price War Against OpenAI and US Rivals

DeepSeek permanently slashed the price of its flagship V4 Pro model by 75%, converting what had been a promotional discount set to expire May 31, 2026 into its permanent pricing tier, Bloomberg and multiple outlets confirmed on May 26. New standard pricing ranges from $0.003625 per million input tokens (cache hits) to $0.87 per million output tokens — making DeepSeek V4 Pro less than one-tenth the cost of OpenAI's GPT-5.5 and meaningfully below Chinese rival Kimi at scale. The price cut deepens a global AI price war that DeepSeek initiated in January 2025 with the R1 model's release, and intensifies pressure on every enterprise AI provider competing for API market share. The timing — as DeepSeek is finalizing its first external funding round at ~$50B valuation — signals that the company has calculated competitive pricing as its primary market penetration strategy rather than near-term margin: consistent with founder Liang Wenfeng's May 22 declaration that the company will prioritize groundbreaking research over commercialization. Caixin Global reported the price cut is directly connected to the funding round's closing timeline, as DeepSeek seeks to maximize API adoption before competition from DeepSeek V5 (anticipated H2 2026) complicates the market position of V4 Pro. The 75% permanent reduction also applies downward pressure on Alibaba's Qwen and Baidu's ERNIE API pricing — both companies have been forced to slash prices repeatedly since January 2025 as DeepSeek's efficiency advantage drives China's domestic AI API prices toward commodity levels. Semafor noted that V4 Pro's pricing is now below the compute cost that Western AI companies estimate for running similar-capability models, reinforcing DeepSeek's training efficiency as a structural rather than a temporary competitive moat.