DeepSeek V4: Chinese company DeepSeek has launched its most powerful AI model so far, V4. The company says that its model will give tough competition to the AI models of American companies Google Deepmind and OpenAI and will perform like them. Two variants of this model have been introduced.
Far ahead of the open source model
The Deekseek V4 AI model is one of the company’s biggest developments to date. Competition is continuously increasing in the global market. In such a situation, the efforts of companies are to bring better and high performance AI models. On an international scale, DeepSeek-V4-Pro is significantly ahead of other open-source models and only slightly behind the closed-source model Google’s Gemini-Pro-3.1, the company said in a statement.
The largest model ever
The V4-Pro model has 1.6 trillion parameters. Due to this, it has become the biggest model of the company so far. The V4-flash model is 284 billion parameters. Both variants support a context window of 1 million tokens. It decides how much information the system can process at one time.
Will work with both Nvidia and Huawei chips
Information about which hardware has been used for training DeepSeek’s V4 model has not been given. However, the company noted that its system includes the ability to work with both Nvidia and Huawei chips. The company further said that costs are expected to decline by the end of the year as new hardware, including Huawei’s Ascend 950PR system, becomes widely available. Now it remains to be seen how true the claims being made by the company about its powerful model will prove to be and whether it will affect Gemini and ChatGPT.
About the authorMona DixitMona Dixit is an experienced journalist in the world of technology and gadgets. She is currently associated with Times Online as the Principal Digital Content Producer. Active in the field of journalism for the last 9 years, Mona has closely observed and covered the changing landscape of technology and consumer electronics. As a tech journalist, Mona specializes in reviews of new gadgets, launch events and changes happening in the world of AI. She not only keeps a keen eye on the latest smartphones, but also deeply examines audio devices and wearables. Mona also has a strong understanding of consumer electronics, AI innovation and the smartphone market. He is an expert in converting complex things of technology into simple life-hacks (Tips & Tricks) and accurately analyzing future technology. Mona believes that the real measure of technology is not just its ‘specifications’ but ‘user experience’. His real skill is to translate complex technical terminology into simple and colloquial language, so that even a common reader can take full advantage of the technology. For them the ultimate purpose of technology is to make life simpler and better. Mona’s real skill is to convey difficult technical information to the readers in easy and colloquial language. He believes that the real purpose of technology is to make life simpler. After Bachelor of Science, Mona has done post graduation in Journalism and Mass Communication from IMS Noida, Meerut University. Before joining Times Online, she had worked with leading technology websites – Techlusive and Newsbytes.… read more