 |
| DeepSeek logo as purple |
DeepSeek Launches V4 Pro With Higher Prices for Advanced AI
BEIJING — August 13, 2026
Chinese artificial intelligence company DeepSeek has officially launched its V4 Pro model, setting prices far above those of its V4 Flash model as it positions the new system as a more powerful premium offering. The company says the new model brings major improvements in tasks that require AI agents to reason, use tools and complete complex work.
DeepSeek’s V4-Pro-0813 is priced at $1.32 per million input tokens and $3.96 per million output tokens, according to independent AI benchmarking company Artificial Analysis. Tokens are small units of text that AI models process when reading or generating information.
The new pricing makes V4 Pro about nine times more expensive than V4 Flash for input and 14 times more expensive for output. Artificial Analysis lists V4 Flash at $0.14 per million input tokens and $0.28 per million output tokens.
DeepSeek said V4-Pro-0813 is available through its application programming interface, or API, as well as its app and web services. The company also plans to change API prices for both V4 Pro and V4 Flash and introduce separate rates for periods of high and low demand.
The higher price places a significant premium on DeepSeek’s flagship model. The move follows an unusual development last month, when the less expensive V4 Flash performed better than an April preview version of V4 Pro in several independent tests.
That result raised questions about the difference between the two models and whether DeepSeek had made rapid improvements to its technology between the preview and the final V4 Pro release. More recent independent testing now indicates that the official V4 Pro has stronger overall performance.
Artificial Analysis gave the reasoning version of V4 Pro a score of 53 on its Intelligence Index, compared with 40 for V4 Flash. The index measures performance across nine areas, including AI-agent tasks, use of external tools, programming, scientific reasoning and the ability to work with large amounts of information.
The launch comes as DeepSeek tries to maintain its position in China’s increasingly competitive AI industry. The company gained international attention after its R1 model became widely discussed in early 2025, prompting debate about whether advanced AI systems could be developed with significantly lower costs than those associated with leading U.S. companies.
Since then, several Chinese competitors have released increasingly capable AI models. Companies including Moonshot AI, Zhipu AI, MiniMax, Alibaba and ByteDance have challenged DeepSeek for attention and market share.
DeepSeek is also facing the challenge of turning its reputation into a sustainable business. The company was reportedly preparing a new fundraising round in July at a valuation of about $74 billion, shortly after securing roughly $7.4 billion in its first external financing round in June.
The move toward outside funding reflects the growing cost of competing in the AI sector. Developing and operating advanced models requires large investments in computing chips, data centres and specialised employees.
DeepSeek has said it plans to at least double its workforce in areas including data-centre operations and AI-agent development. The company has also increased private recruitment of chip-design engineers as it works on its own AI processor, an effort that could eventually reduce its dependence on suppliers such as Nvidia and Huawei.
With V4 Pro, DeepSeek is seeking to turn stronger technical performance into a higher-value commercial product. The model’s success will depend not only on benchmark results but also on whether customers are willing to pay substantially more for its additional capabilities.
.png)
.png)
