© 2026 NervNow™. All rights reserved.
ByNN Desk
DeepSeek’s New V4 Model Fails to Close the AI Gap
China's DeepSeek released its long-awaited V4 model on April 24, 2026, yet it still trails leading US AI models by an estimated 3 to 6 months.
China’s DeepSeek released its long-awaited V4 model on April 24, 2026, yet it still trails leading US AI models by an estimated 3 to 6 months.
More than a year after DeepSeek released its competitive R1 model, which was purportedly built for less than many rivals, the company has now unveiled preview versions of its long-awaited new flagship model, called V4. However, as experts confirm, the latest release does not meaningfully shift the balance of power in the global AI race.
The Hangzhou-based startup released the model in two versions: DeepSeek V4-Pro, a larger model aimed at more demanding tasks, and DeepSeek V4-Flash, a smaller version designed to respond faster and cost less to run. Furthermore, both models are mixture-of-experts models with context windows of 1 million tokens each enough to allow large codebases or documents to be used in prompts.
In terms of raw size, the Pro model has a total of 1.6 trillion parameters (49 billion active), making it the biggest open-weight model currently available a notable technical milestone.
Despite the impressive specifications, the benchmarks tell a more nuanced story. DeepSeek’s own accompanying research paper conceded that, in certain respects, the new model lags the most cutting-edge AI software in the US.
Specifically, DeepSeek’s tech report says the V4 “falls marginally short of GPT-5.4 and Gemini 3.1 Pro, suggesting a developmental trajectory that trails state-of-the-art frontier models by approximately three to six months.
Moreover, the timing of the release made things worse for DeepSeek. It’s model is likely even further behind OpenAI’s GPT-5.5, a newer offering released the day before V4 that’s designed to be better at completing tasks without much direction.
Nevertheless, DeepSeek is not without its strengths and price is arguably its most powerful one. DeepSeek’s V4-Pro model costs $3.48 per 1 million output tokens; by comparison, OpenAI and Anthropic charge $30 and $25 respectively for the same volume of work.
Beyond performance and pricing, another development deserves attention. To fulfill V4’s computing needs, DeepSeek partnered with Chinese tech giant Huawei, which supports the AI startup with its Supernode technology by combining large clusters of its Ascend 950 chips to provide more computing power.
This is a significant departure from its predecessor. Another big change is they are powered by computer chips made by Huawei, not US chipmakers like Nvidia. Analysts note this reduces China’s dependence on US semiconductor exports a key strategic goal amid ongoing export restrictions.
The launch, however, has not come without controversy. The release came a day after the White House accused China of stealing American AI labs’ IP on an industrial scale using thousands of proxy accounts.
Additionally, in February, Anthropic accused DeepSeek and two other China-based AI laboratories of industrial-scale campaigns to illicitly extract Claude’s capabilities to improve their own models, using a technique called distillation that involves training a less capable model on the outputs of a stronger one. OpenAI made similar allegations in a letter to US lawmakers.
Unlike the shock of R1, the market response this time appears calm. R1 shocked US markets because no one expected a Chinese model to compete at that level. V4 is simply a follow-through on that same trend, and trends don’t make headlines the way shocks do,” said Ivan Su, senior equity analyst at Morningstar.
ALSO READ : Verda Raises $117M to Meet Surging Demand for AI Infrastructure
That said, analysts acknowledge DeepSeek remains a force to reckon with in the longer term. Industry experts noted that DeepSeek’s pricing has reset the developmental trajectory of the field, placing massive pressure on closed-source providers like OpenAI and Anthropic to justify their premiums.
DeepSeek’s prices could get even cheaper: It expects to lower V4-Pro prices later in the year as Huawei scales up production of its new Ascend 950 AI processors.
Furthermore, the existing deepseek-chat and deepseek-reasoner endpoints will be fully retired on July 24, 2026, with all traffic rerouted to the V4-Flash architecture.
In summary, DeepSeek V4 is a technically capable and commercially disruptive model, but one that ultimately confirms, rather than challenges, the United States’ lead in frontier AI development. Whether that changes with the final, non-preview release remains the question the industry will be watching closely.
Disclaimer: This news is based on publicly available information. NervNow has not independently verified any claims.
MORE ON AI
Anthropic Hits $1 Trillion Mark as Tech Giants Compete Fiercely
South Korea’s Naver Signs AI and Cloud MoU with TCS
Google in Talks With Marvell to Design Two New AI Inference Chips
OpenText Brings AI and Data Solutions to AWS European Sovereign Cloud
Globant and Autodesk Tandem Join Forces to Advance Physical AI
