Key Points:
- Chinese AI startup DeepSeek has made its massive 75% price discount on the flagship DeepSeek-V4-Pro model permanent.
- Under the locked-in rates, input costs for cache misses dropped to $0.435 per million tokens, while output costs fell to $0.87 per million tokens.
- The permanent price drop arrives as Western developers face tightening rate limits and usage restrictions on US-based AI platforms.
- The Hangzhou-based firm is currently seeking new funding at a valuation of roughly $45 billion to prioritize open-source research.
Chinese artificial intelligence laboratory DeepSeek has made its massive 75% discount on its flagship DeepSeek-V4-Pro model permanent. The Hangzhou-based company initially introduced the steep price cut as a temporary promotional campaign in late April, set to expire on May 31, 2026. However, the startup decided to lock in the heavily reduced rates permanently to attract developers and businesses looking for highly cost-effective alternatives to Western AI platforms.
The permanent price adjustment establishes DeepSeek-V4-Pro as one of the most affordable frontier AI APIs on the market. Under the revised pricing structure, the cost for non-cached input tokens (cache misses) dropped from $1.74 to just $0.435 per million tokens. Meanwhile, input cache hits fell from $0.145 to a mere $0.03625 per million tokens. For output tokens, developers now pay only $0.87 per million tokens, down from the original $3.48. These changes put DeepSeek’s operational costs at a fraction of those charged by major competitors.
DeepSeek’s aggressive pricing strategy directly targets growing frustration within developer communities regarding Western AI platforms. Over the last few months, users of popular US consumer chatbots and APIs have reported hitting sudden, restrictive usage caps. Major players like Google, Anthropic, and OpenAI have quietly tightened user gateways through complex monitoring systems to manage the soaring infrastructure costs of advanced computing. DeepSeek is positioning itself as a stable and unrestricted haven for creators who want to run large-scale autonomous agents without fear of sudden lockouts.
This pricing move highlights a broader divergence in how global AI hubs approach technology deployment. Industry analysts note that Asian AI platforms are successfully shifting toward a token-based economy, leveraging low electricity costs and a massive developer pool. By treating AI processing units as tradable, low-cost assets, Chinese laboratories can afford to run at margins that American tech firms struggle to match. This industrial approach has fueled a surge in productivity, particularly among small startups and independent software developers.
DeepSeek first shook up the global technology sector in 2025 with the release of its R1 model. That open-weight model demonstrated reasoning capabilities comparable to OpenAI’s top platforms but cost only a tiny fraction of the billions of dollars spent by Silicon Valley competitors. The release of R1 temporarily triggered a trillion-dollar selloff in global tech stocks as investors questioned the necessity of massive infrastructure outlays. Despite those brief jitters, capital expenditure has rebounded, with US tech giants planning to invest around $650 billion in AI data centers in 2026 alone.
To fund its rapid expansion and research, DeepSeek is currently finalizing a major fundraising round. The startup has held discussions with potential investors about a 70 billion yuan ($9.7 billion) funding round, which would value it at approximately $45 billion. Despite this massive valuation, founder Liang Wenfeng has reportedly assured investors that the company will prioritize groundbreaking open-source research over near-term commercialization and monetization.
The company’s decision to stick with open-source systems aligns with its stated goal of achieving artificial general intelligence (AGI). Zhejiang High-Flyer Asset Management, a prominent quantitative hedge fund, owns and backs DeepSeek. Because High-Flyer underwrites the operational costs from its hedge fund profits, the AI lab operates without the typical pressure from external venture capitalists demanding immediate returns. This unique financial setup allows DeepSeek to run aggressive price wars, keeping its flagship API pricing locked at rock-bottom rates.
