China’s DeepSeek Finally Launches A New AI Model

More than a year after shocking the world with a low-cost reasoning model that matched the powers of US rivals, Chinese startup DeepSeek unveiled a new AI model on Friday.

The Chinese AI business DeepSeek formally unveiled a preview of their eagerly anticipated DeepSeek-V4 model earlier today. The R1 reasoning model, which disrupted the worldwide market by matching Western AI capabilities for a tenth of the cost, had a huge influence prior to its release. 

The AI company had praised DeepSeek-V4 as “cost-effective” in a separate announcement on X and said that it “features an ultra-long context of one million words” in a statement on social media platform WeChat.

The statement coincided with Meta’s revelation that it would lay off 10% of its workforce in an effort to increase productivity while making significant investments in artificial intelligence. Microsoft reportedly wanted to reduce its workforce as well. 

The context length of DeepSeek-V4 “(achieves) leadership in both domestic and open-source fields across agent capabilities, world knowledge, and reasoning performance,” which establishes how much input a model can take in to help it finish tasks seamlessly.

The business states that the open source concept is now accessible in a “preview version.”

Two versions of DeepSeek-V4 are available: DeepSeek-V4-Pro and DeepSeek-V4-Flash. The latter is “a more efficient and economical choice” due to its smaller parameters.

The models’ ability to make decisions is improved by the V4-Pro’s 1.6 trillion parameters and the V4-Flash’s 284 billion parameters.

In what was sometimes referred to as a “Sputnik moment” for the sector, this alleged “DeepSeek shock” has caused a sell-off of AI-related stock and a re-evaluation of corporate strategy.

Comparing the chatbot’s performance with that of ChatGPT and other leading American products, the company claimed that it required a lot less processing power to create.

However, the chatbot frequently declined to respond to inquiries about delicate subjects like the 1989 Tiananmen crackdown, raising concerns about data privacy and censorship. 

Chinese towns, healthcare facilities, the financial industry, and other enterprises have all embraced DeepSeek’s AI tools.

This has been influenced in part by DeepSeek’s choice to make its systems publicly available and open source, as opposed to the proprietary models offered by OpenAI and other Western competitors.

Chinese Premier Li Qiang said last month at an annual meeting of China’s top decision-makers that “China-made large AI models spearheaded the development of the global open-source AI ecosystem.”

The rivalry between China and the US has escalated as a result of the AI race, and the White House accused Chinese organizations on Thursday of making a significant attempt to steal AI technology.

In an X post, science and technology head Michael Kratsios stated, “The US has evidence that foreign entities, primarily in China, are running industrial-scale distillation campaigns to steal American AI.”

“We are going to take steps to safeguard American innovation.”

On the other hand, DeepSeek-V4 features a massive one-million-word context window for large-scale analysis, an open-source codebase for local modification, native support for AI agent tools like Claude Code and OpenClaw to automate complex tasks, and specialized variants including a high-performance “Pro” version and a faster, more affordable “Flash” option.

DeepSeek claims V4 drastically reduces compute and memory costs, with analysts noting world-class coding at lower inference costs than U.S. rivals, while Huawei’s full-stack support for V4 on its Ascend AI processors marks a milestone for domestic hardware integration and reduced reliance on restricted Nvidia chips, and despite a slight price increase, the V4-Pro remains about seven times cheaper than Anthropic’s Claude Opus 4.6.