DeepSeek Debuts Upgrade to AI Model That Improves Reasoning and Coding
DeepSeek introduced an upgrade to its artificial intelligence model.
The new version of the Chinese startup’s V3 large language model was made available through AI development platform Hugging Face, Reuters reported Tuesday (March 25).
The release marks DeepSeek’s latest effort to make a name for itself in an evolving AI sector, competing with the likes of OpenAI and Anthropic, the report said. The new model shows improvements over its predecessor in areas like reasoning and coding.
DeepSeek rocked the tech world earlier this year when it unveiled a series of AI models that were said to perform at the same level as OpenAI’s ChatGPT but at a lower cost.
Interviewed at a conference in China this week, Apple CEO Tim Cook reportedly described DeepSeek’s AI models as “excellent.”
Asked about the risk of DeepSeek during Apple’s earnings call Jan. 30, Cook said: “In general, I think innovation that drives efficiency is a good thing. And that’s what you see in that model.”
But while the debut of DeepSeek led observers to question the need for investment in AI infrastructure, it also prompted a greater focus on reasoning models, which require more spending on inference.
AI investments by hyperscale companies such as Amazon, Meta and Microsoft will increase faster than earlier forecasts, with more of that money being spent on running AI systems once they have been trained, instead of on data centers and chips, Bloomberg Intelligence found.
The data showed those companies spending $371 billion on data centers and computing resources in 2025 — 44% more than they spent last year — and $525 billion a year by 2032.
Meanwhile, AI is being used to speed up the work of customer experience teams and make customers feel heard and better understood.
“We’re finally moving beyond superficial ‘personalization,’” said Lisa O’Malley, senior director of industry products and solutions at Google Cloud, in a blog post last month. “AI-powered CX creates the feeling of being understood, of having needs anticipated and met with minimal effort.”
For example, O’Malley said customers have begun saying “please” and “thank you” to AI agents, although the “most significant shift, however, is the evolution of the support system from a cost center to a revenue generator. The conversations I’m having with customers point to omnichannel engagement — across voice, web, mobile, email and apps — as directly driving ROI.”
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
The post DeepSeek Debuts Upgrade to AI Model That Improves Reasoning and Coding appeared first on PYMNTS.com.