DeepSeek Makes 75% Price Cut Permanent — AI Price War Enters Most Aggressive Phase
DeepSeek has made its 75% price reduction on the V4-Pro model permanent as of May 24, 2026, signaling that the AI pricing war has entered its most aggressive phase yet. The move means developers and enterprises can now access frontier-class reasoning capabilities at roughly one-quarter the cost of comparable tokens from OpenAI or Anthropic.
The strategic timing is notable. Chinese AI models already account for 60% of all usage on OpenRouter, the multi-model inference platform, with DeepSeek being the single largest contributor to that share. By locking in the discount permanently, DeepSeek is effectively attempting to cement market share before a wave of new model releases from US labs later this summer.
For builders and startups in MENA and globally, the implications are immediate. Token costs — one of the largest line items for AI-powered products — just dropped substantially, potentially enabling new product categories that were previously uneconomical. The move also puts pressure on OpenAI, Anthropic, and Google to respond with their own pricing adjustments within the coming weeks.
Industry analysts expect this to trigger a cascade of quiet pricing changes across all major AI providers. The question is no longer whether prices will fall further, but how fast labs can maintain margins while competing with Chinese pricing.
This is the most consequential pricing move in AI since DeepSeek's original launch disrupted the market. A permanent floor price at 75% below Western alternatives fundamentally changes build-vs-buy economics and forces every AI company to justify their premium.
Why did DeepSeek make the 75% price cut permanent?
DeepSeek is locking in market share before US labs release new models this summer. Chinese models already represent 60% of OpenRouter usage, and a permanent price cut discourages developers from switching back.
How does this affect AI developers and startups?
Token costs drop to roughly one-quarter of Western alternatives, making previously uneconomical AI product categories viable. Expect competing providers to adjust pricing within weeks.