DeepSeek Launches V4 Model Amid Controversy and Competition

On April 24, 2026, DeepSeek, a Chinese artificial intelligence startup, unveiled a preview of its new foundational AI model, the V4, which is designed to compete with leading closed-source models from companies like OpenAI and Google DeepMind. This launch marks a significant shift for DeepSeek, which previously relied on Nvidia chips for its models. The V4 model is notable for its two versions: the V4-pro, featuring 1.6 trillion parameters, and the V4-flash, with 284 billion parameters. Both versions boast a context window of 1 million tokens, significantly enhancing their processing capabilities compared to DeepSeek's earlier models.

South China Morning PostReutersCNN

DeepSeek's collaboration with Huawei is central to the development of the V4 model. Huawei's Ascend AI chips are now fully integrated with DeepSeek's V4 series, allowing the models to run efficiently on Huawei's high-performance systems. This partnership reflects China's broader strategy to achieve technological self-sufficiency in the face of U.S. export restrictions on advanced AI chips.

Reuters

The launch of the V4 model comes at a time when the U.S. government has accused China of engaging in industrial-scale intellectual property theft related to AI technologies. DeepSeek has been implicated in these allegations, with claims that it improperly utilized proprietary models from companies like OpenAI and Anthropic. Despite these accusations, DeepSeek asserts that its models are built on data collected through web crawling and not on synthetic data generated by competitors.

Reuters

The introduction of the V4 model has already affected the stock market, with shares of rival companies such as Zhipu AI and MiniMax experiencing declines following the announcement. Analysts suggest that while the V4 model is a continuation of DeepSeek's competitive edge, it may not generate the same market frenzy as its predecessor, the R1 model, which shocked investors with its unexpected performance.

CNN

The Chinese Embassy in Washington has dismissed the allegations against DeepSeek as "baseless," emphasizing China's commitment to intellectual property rights. Huawei has expressed its support for DeepSeek, stating that its Ascend chip line is crucial for the startup's new models. Wei Sun, a principal analyst at Counterpoint Research, noted that the V4's reliance on domestic chips could enhance its impact on both the Chinese market and global AI development.

Reuters

Despite its advancements, DeepSeek faces skepticism from industry leaders. OpenAI and Anthropic have accused the startup of illegally extracting capabilities from their models, raising concerns about the ethical implications of its rapid rise. The ongoing tensions between the U.S. and China further complicate the narrative, with U.S. officials warning of potential risks associated with Chinese AI developments.

CNN

As DeepSeek prepares for the final launch of the V4 model, the company aims to raise funds at a valuation exceeding $20 billion. The final product is expected to further solidify DeepSeek's position in the AI landscape, while the broader implications of its development will continue to unfold amid geopolitical tensions and competition in the tech sector.

South China Morning Post