Chinese company DeepSeek releases long-awaited new version of its AI system
DeepSeek, a Chinese start-up, has released updated versions of its artificial intelligence chatbot: DeepSeek-V4-Pro and DeepSeek-V4-Flash. These models follow an open source approach, allowing developers to freely use and adapt the code.
DeepSeek said V4-Pro outperforms all other open source models in terms of mathematical and programming skills. Only Google’s closed-source Gemini 3.1-Pro does better when it comes to general knowledge. The performance of the ‘pro’ version is almost on a par with OpenAI’s GPT‑5.4 and Gemini 3.1-Pro. That points to a development pace that lags up to six months behind the leading models.
Meanwhile, V4-Flash offers similar reasoning capabilities to V4-Pro, but with faster response times and lower usage costs. This follows the impressive launch of DeepSeek-R1 in January last year. At the time, analysts compared that model’s capabilities to ChatGPT and Gemini.
Venture capitalist Marc Andreessen described the introduction of DeepSeek-R1 as a pivotal moment for AI. Its performance was particularly striking. DeepSeek claimed the model cost less than $6 million in computing expenses, a sharp contrast with the multibillion-dollar budgets that are standard in Silicon Valley.
Some analysts questioned those figures. They suggested that DeepSeek had access to more funding and advanced chips. Even so, the arrival of the start-up undoubtedly turned the technology landscape upside down. Concerns about data protection and possible censorship by the Chinese government led several countries to impose bans or restrictions on DeepSeek-R1. Among them were the US, Australia, Taiwan, South Korea, Denmark and Italy, all citing risks to privacy and national security.
Business AM
Subscribers 0
Fans 0
Followers 0
Followers