- V4 Pro model launch and pricing updates
- Performance evolution and superiority of previous Flash models
- Massive fundraising rounds and record valuation
- Doubling workforce and manufacturing custom chips
- Intense competition in the artificial intelligence market
- Frequently asked questions
V4 Pro model launch and pricing updates
Beijing-based Chinese artificial intelligence startup DeepSeek has officially launched its updated flagship V4 Pro model in a strategic move aimed at strengthening its leading position amid rising competition from domestic and international tech companies. Released under the technical designation V4 Pro 0813, the model focuses fundamentally on delivering a qualitative leap in the capabilities of intelligent agents designed for complex tasks and autonomous planning across various platforms.
The company confirmed that the new model is now available for immediate use via application programming interfaces (APIs), the dedicated smart device app, and the official web portal. This rollout was accompanied by DeepSeek’s announcement of a revised API pricing structure for both the V4 Pro and V4 Flash models, introducing dynamic pricing that differentiates between peak usage periods and off-peak hours in order to manage server loads flexibly and optimize operational returns for the computing infrastructure.
Performance evolution and superiority of previous Flash models
This launch has drawn exceptional attention and close monitoring from industry experts and researchers, particularly following the surprise generated earlier by the lower-cost V4 Flash model, which unexpectedly outperformed the initial experimental version of V4 Pro released last April in several independent benchmark tests. Observers viewed those results as clear evidence of the rapid pace of development undertaken by DeepSeek’s engineering teams between the experimental preview stage and the final approved release.
The company has demonstrated superior capability in algorithmic optimization and efficient model training, enabling it to bridge technical gaps and deliver a final version capable of handling complex programming contexts and analytical logic at accuracy levels matching the latest developments from Silicon Valley research labs.
Massive fundraising rounds and record valuation
DeepSeek has emerged as one of the most influential and controversial AI companies globally since early 2025, following the renowned launch of the R1 model, which reshaped global discussions on the feasibility of building high-capability AI systems at significantly lower computing and training costs compared to the massive expenditures incurred by major US firms. However, the company now faces the complex challenge of translating its fame and technical innovations into a long-term sustainable and profitable business model.
In this context, informed press reports in July revealed that DeepSeek is planning a new funding round that could push its market valuation to nearly $74 billion, just weeks after closing its first external funding round in June valued at approximately $7.4 billion. This trend represents a radical shift in the company’s philosophy, which long relied on self-funding, highlighting the escalating costs of the technological race and its massive requirements for processing chips, data centers, and power sources.
Doubling workforce and manufacturing custom chips
To keep pace with this expansion, DeepSeek announced an ambitious strategy aimed at doubling its headcount across various departments, with a focus on attracting top researchers in data center architecture, smart agent development teams, and systems security experts. These expansions come as the company seeks to build an integrated software environment that supports ultra-large-scale training and inference operations.
On the hardware front, the company has intensified its covert moves to hire integrated circuit and chip design engineers to develop its own proprietary AI chip. Through this bold step, DeepSeek aims to reduce its reliance on external suppliers such as Nvidia and Huawei, securing stable supply chains for its computing needs that protect it from geopolitical constraints and global market volatility.
Intense competition in the artificial intelligence market
DeepSeek faces intense competitive pressures within the Chinese market as startups and giants alike—such as Moonshot AI, Zhipu AI, and MiniMax, alongside Alibaba and ByteDance—race to roll out advanced smart models at competitive prices targeting developers and business sectors. This heated landscape compels DeepSeek to maintain rapid innovation in model architecture and provide easily integrated software tools for global enterprises.
The launch of V4 Pro underscores that the AI battle is entering a new phase focused on the practical applications of autonomous agents capable of decision-making and complex problem-solving, as DeepSeek seeks to solidify its position as a trusted provider of smart infrastructure capable of combining technical efficiency with well-calculated economic cost.
Frequently asked questions
Q: What channels are available to access the new DeepSeek V4 Pro model?
A: The model is available via developer APIs, the smartphone application, and the official website.
Q: What is the reason behind DeepSeek’s adjustment of API pricing?
A: Prices were adjusted for the V4 Pro and Flash models, and peak/off-peak pricing was introduced to manage server pressure and cover increasing computing costs.
Q: What are DeepSeek’s plans in electronic chip manufacturing?
A: The company has begun recruiting chip design engineers to build its own AI processors to reduce reliance on Nvidia and Huawei.
