DeepSeek-V4-Pro Makes Permanent 75% Price Cut — Chinese Model "At One-Quarter the Price"Shakes Up Global API Market, Chinese Models Dominate 60% of OpenRouter Share

5.25 While Western frontier labs are still struggling with trillion-dollar valuations and profitability, DeepSeek has proven an alternative path through extreme cost-performance: leveraging open-source, low-cost pricing, and a multi-model matrix to rapidly capture the developer ecosystem, then using ecosystem scale to feed back into technological iteration.

On the evening of May 22, DeepSeek officially announced that the API price of its DeepSeek-V4-Pro model will be permanently adjusted to one-quarter of the original pricing after the 25% discount promotion ends on May 31—meaning the previous limited-time offer is transitioning from a short-term promotional tactic to normalized pricing. This decision sent shockwaves through the global AI model pricing market, arriving at a critical juncture as Anthropic races toward a $90 billion valuation and OpenAI prepares for a trillion-dollar IPO.

The Numbers Speak: Chinese Models Already Dominate 60% of OpenRouter Share

The latest data from OpenRouter, the world's most widely used third-party AI model router, shows that Chinese models now account for 60% of total usage. DeepSeek's models recorded a weekly total of 5.74 trillion tokens in calls, up 25.9% month-over-month, surpassing Anthropic and Google to rank first for two consecutive weeks.

Global weekly large model call volumes have risen for five consecutive weeks, reaching a total of 28.9 trillion tokens. Notably, Moonshot AI's Kimi K2.6, previously ranked sixth, has dropped off the list, while three DeepSeek models—V4-Flash, V4-Pro, and V3.2—have simultaneously entered the top nine, forming a multi-model matrix strategy to acquire users across different segments.

The "Asymmetric Strike" of the Price War

DeepSeek's price reduction strategy is delivering an "asymmetric strike" effect:

DimensionDeepSeek StrategyWestern Vendor Dilemma
PricingV4-Pro permanently at 25% of original priceClaude's cost is 9x that of DeepSeek (same benchmark)
TechnologyMillion-token context + open-source ecosystemClosed-source models struggle to match price cuts
Market60% OpenRouter shareHigh consumer-side price sensitivity
EcosystemMulti-model matrix covering diverse scenariosSingle-flagship model strategy is constrained

Industry Impact: A Turning Point from "Price War" to "Value War"

DeepSeek's permanent price cut is not merely a battle for market share, but a fundamental challenge to the global AI business model. When high-performance models can be offered at ultra-low prices on a normalized basis, Western vendors' high-margin API business models face a systemic shock.

According to the latest data from the National Data Bureau, as of March 2026, China's daily token call volume has exceeded 140 trillion, a more than 1,000-fold increase from the 100 billion recorded at the start of 2024. Tokens have become the value anchor and commercial settlement unit of the intelligent era.

Conclusion

DeepSeek's "one-quarter pricing" is not a promotion—it is a declaration. It signals that Chinese AI models have transitioned from "challengers" to "rule-makers." While Western frontier labs are still struggling with trillion-dollar valuations and profitability, DeepSeek has proven an alternative path through extreme cost-performance: leveraging open-source, low-cost pricing, and a multi-model matrix to rapidly capture the developer ecosystem, then using ecosystem scale to feed back into technological iteration. This "encircling the cities from the countryside" strategy is quietly shifting the center of gravity of global AI pricing power eastward. Anthropic's $90 billion valuation is built on a narrative of "enterprise-grade high margins," but DeepSeek's 60% OpenRouter share is proving that in the price-sensitive developer and SME market, "good enough and cheap enough" may be more disruptive than "the best but the most expensive."