DeepSeek has slashed V4-Pro API prices by 75% through May 5, bringing output to $0.87 per million tokens and cache-hit input to $0.0036 per million—roughly 14–29x cheaper than comparable Western frontier models priced at $12–25 per million tokens for output. Cache-hit input rates across the broader DeepSeek model family have been cut to one-tenth of standard pricing. The move deepens the Chinese AI price war, where domestic labs are aggressively undercutting to accelerate developer adoption and compete globally with the V4-Pro model released last week.