SHANGHAI (TECH/BUSINESS DESK, AUG 14): DeepSeek’s latest flagship AI model debuts with significantly higher pricing than V4 Flash, reflecting enhanced reasoning capabilities, stronger benchmark performance and the company’s push to strengthen its commercial position amid intensifying competition.
DeepSeek has officially introduced its V4 Pro artificial intelligence model, positioning it as a premium offering with prices that are substantially higher than its V4 Flash model as the Chinese AI startup looks to monetize improvements in performance and advanced capabilities.
The newly released V4-Pro-0813 model is priced at $1.32 per million input tokens and $3.96 per million output tokens, according to benchmarking firm Artificial Analysis.
Those rates are roughly nine times higher for input tokens and 14 times higher for output tokens than V4 Flash, which is listed at $0.14 per million input tokens and $0.28 per million output tokens.
DeepSeek said V4-Pro-0813 delivers major improvements for AI agent applications and is available across its API, mobile application and web platform. The company also announced revised API pricing for both V4 Pro and V4 Flash, introducing peak and off-peak pricing structures.
The premium pricing reinforces DeepSeek’s strategy of clearly separating its flagship model from V4 Flash. Last month’s release of V4 Flash surprised industry observers after it outperformed an April preview version of V4 Pro in several independent benchmark tests despite being positioned as the lower-tier model.
That unexpected result suggested DeepSeek had rapidly advanced its underlying technology between the preview and official releases.
Recent benchmark results indicate the finalized V4 Pro has made a notable leap in performance. Artificial Analysis assigned the reasoning version of V4 Pro an Intelligence Index score of 53, compared with 40 for V4 Flash.
The Intelligence Index measures performance across nine categories, including agentic work tasks, tool usage, coding, scientific reasoning and long-context understanding.
DeepSeek first gained global attention after its R1 model surged in popularity in early 2025, fueling debate over whether advanced AI systems could be developed at significantly lower costs than competing models from U.S. companies.
Since then, however, the company’s early momentum has been challenged by a wave of competing AI releases from Chinese firms including Moonshot AI, Zhipu AI, MiniMax, Alibaba and ByteDance.
Alongside technological competition, DeepSeek is also working to strengthen its commercial foundation. The company is reportedly preparing a new fundraising round that could value it at around $74 billion, following its first external financing round in June that raised approximately $7.4 billion.
The move represents a shift for DeepSeek, which had previously operated without outside investment. Rising infrastructure costs, including spending on AI chips, data centres and specialized talent, have increased the financial demands of competing in the rapidly evolving AI industry.
DeepSeek has also said it plans to at least double its workforce across multiple divisions, including AI agent development and data-centre operations. In addition, the company has expanded recruitment of chip-design engineers as it works on developing its own AI processors, a strategy aimed at reducing dependence on suppliers such as Nvidia and Huawei.
DeepSeek’s latest pricing and product strategy underscores its ambition to position V4 Pro as a premium AI platform while investing heavily to remain competitive in the global race for advanced artificial intelligence.
Also read: Bank of Japan Signals September Rate Hike as Officials Weigh Faster Pace of Monetary Tightening
