DeepSeek price slash fuels competition, hits Zhipu, Minimax

Chinese artificial intelligence startup DeepSeek has introduced a preview of its fourth-generation model, DeepSeek V4, alongside sweeping price cuts aimed at accelerating adoption.

The company said it is offering developers a 75% discount on its newly unveiled AI model, DeepSeek-V4-Pro, until May 5.

It also reduced prices for input cache hits across its entire API lineup to one-tenth of the original cost, according to a post on X.

The aggressive pricing move comes as competition intensifies among AI developers, particularly in China, where firms are racing to gain market share.

Two-tier model strategy targets broader user base

DeepSeek said V4 is being introduced in two versions a more powerful Pro model and a lighter, cheaper Flash variant.

The company described the Pro version as outperforming other open-source models in world-knowledge benchmarks, trailing only Google’s closed-source Gemini-Pro-3.1.

According to the company, the V4 models are particularly suited for AI agent tasks, which involve executing more complex operations than traditional chatbots but require greater computing power.

The preview marks DeepSeek’s most significant update since the launch of its V3 model in December 2024.

The company stated that V4 will succeed V3 and represents a major upgrade.

DeepSeek did not disclose detailed pricing beyond the discounts, full performance benchmarks, or a final release timeline.

Instead, it described the preview as an early-stage rollout aimed at gathering real-world feedback before a broader launch.

Price cuts raise concerns over AI sector competition

DeepSeek’s pricing strategy has raised concerns about a potential price war in the AI industry.

API pricing and usage costs have become a key battleground as companies attempt to attract developers and scale adoption beyond niche applications.

Shares of Chinese AI firms reacted negatively to the development.

Zhipu AI, trading as Knowledge Atlas Tech Joint Stock (HK:2513), fell 3.4% in Hong Kong, while MiniMax Group Inc (HK:0100) dropped 10%, extending losses from the previous session.

Both companies had already declined about 9% on Friday following the V4 announcement, amid concerns that DeepSeek’s model could rival closed-source alternatives while using less computing power.

Huawei chip link adds strategic significance

DeepSeek’s V4 model is also drawing attention due to its reported adaptation for Huawei’s chip technology.

According to a report by The Information, the model is expected to run on Huawei’s latest chips.

This development is being closely watched as Chinese AI firms adjust to US export controls and seek domestic alternatives to Nvidia hardware.

The rollout is seen not only as a product launch but also as a test of China’s ability to scale advanced AI models using locally sourced infrastructure.

Market interest intensifies amid AI race

The launch comes as DeepSeek attempts to build on the momentum generated by its V3 model, which gained global attention for delivering strong performance at relatively low cost.

The company has positioned V4 as a superior successor, signalling its intent to strengthen its standing in the rapidly evolving AI landscape.

Meanwhile, competition continues to grow.

Chinese internet giant Tencent Holdings Ltd also unveiled its flagship AI model, Hy3, developed under a former OpenAI researcher, adding further pressure to the sector.

With global and domestic rivals accelerating innovation, DeepSeek’s latest move is likely to intensify scrutiny on its execution and long-term strategy, as investors and developers closely track the next phase of the AI race.