China’s Brahmastra DeepSeek V4 launched in the global tech war. Is the reign of Google and OpenAI over now?

News India Live, Digital Desk: China has taken such a leap in the world of Artificial Intelligence (AI) that has created a stir in the corridors of Silicon Valley (America). Chinese AI Startup DeepSeek Has launched its most powerful model on Friday (April 24, 2026) ‘DeepSeek V4’ Has been launched. This launch is not only technically important, but on the geopolitical front it is also being considered a big victory for China against America. It is being claimed that this model is not only giving tough competition to Google’s ‘Gemini’ and OpenAI’s ‘GPT-5’ level models, but is also better than them in many ways.

V4 becomes ‘superpower’ with 1.6 trillion parameters

DeepSeek V4 is offered in two main versions:

DeepSeek-V4-Pro: This is the heaviest model of the company, in which 1.6 trillion parameters Are. It has been included in the world’s top-3 models in coding, mathematical calculations and logical thinking.

DeepSeek-V4-Flash: It is a low-budget and high-speed model, specially designed for everyday chatbots and AI agents.

Miracle of ‘indigenous’ chip amid US sanctions

The biggest discussion of this launch is not its technology, but the ‘hardware’ that runs it. The US has imposed strict restrictions on Nvidia’s high-end AI chips supplied to China. In such a situation, instead of accepting defeat, China Huawei’s ‘Ascend 950’ chips And this model has been prepared using ‘SuperPoD’ technology. This proves that China is no longer dependent on American hardware for AI.

Why ‘alarm bells’ for Google and OpenAI?

The magic of open source: While the most powerful models from Google and OpenAI are closed-source and cost a fortune, DeepSeek V4 open-source Is. That means developers around the world can use it for free or at very low cost.

Huge 85% reduction in cost: According to reports, the cost of using DeepSeek V4 is closer to that of GPT-5.5. 85% less Is. It is so cheap that even middle class companies can easily adopt it.

Capacity of 10 lakh tokens: This model can process data of up to 1 million words (tokens) at a time, which is equivalent to Google Gemini. That means it can understand entire books and hours of videos in seconds.

‘Show of strength’ just before Trump-Jinping meeting

Experts from the technology world believe that the timing of this launch is very precise. By presenting this model just before the meeting of Trump and Jinping to be held next month, China has given the message that it now has an equal footing in the ‘tech-war’. The arrival of DeepSeek V4 will further strengthen China’s threat in the global market amidst the recent allegations of ‘digital piracy’ made by the White House.