DeepSeek Launches AI Model Upgrade Amid OpenAI Rivalry—Here’s What To Know

/ DeepSeek / By

DeepSeek released an upgrade to its large language model this week, an update the company said featured “significant improvements” over its predecessor as the China-based startup appeared to escalate its rivalry with OpenAI and other U.S. artificial intelligence firms, after an earlier release rattled global tech stocks.

DeepSeek launched an upgrade to its V3 large language model, DeepSeek-V3-0324, on the AI development platform Hugging Face on Tuesday, which the startup marketed as including improvements in reasoning and coding capabilities over its earlier V3 model.

DeepSeek’s V3 large language model was released on Dec. 26, 2024, and the company claimed it was three times faster than its V2 model as it ranked among the top 10 in Chatbot Arena, a performance platform hosted by the University of California, Berkeley.

The latest update included “significant improvements” across several benchmark tests for the language model, DeepSeek said, in addition to upgrades to front-end web development, Chinese writing proficiency—with new features like “interactive rewriting,” according to the company—and Chinese search capabilities like “enhanced report analysis.”

Get Forbes Breaking News Text Alerts: We’re launching text message alerts so you’ll always know the biggest stories shaping the day’s headlines. Text “Alerts” to (201) 335-0739 or sign up here.

What To Watch For

Whether the latest DeepSeek model impacts U.S. tech stocks. Shares for Nvidia dropped by 1.2% as of 9:35 a.m. EDT, while other stocks like Broadcom (0.6%) and Tesla (0.3%) declined slightly. Apple (0.3%), Meta (0.7%) and Microsoft (0.3%) were up Tuesday morning.

What Is Deepseek?

DeepSeek is an AI startup founded by Chinese entrepreneur Liang Wenfeng in 2023. Liang started accumulating thousands of Nvidia graphics processors for an unnamed AI project in 2021, just before the Biden administration restricted trade of those chips to China. He called for China to “gradually transition” to being a contributor in the AI industry, “rather than continuing to ride on the coattails of others.” DeepSeek has claimed its products are more efficient and cost less to train and develop compared to similar products made by OpenAI and Meta. The company said training one of its latest models cost about $5.6 million, far less than the $100 million to $1 billion an AI executive estimated for costs to build a model.

Will Deepseek Be Banned In The U.s.?

A bipartisan bill banning Deepseek from federal devices was introduced in February, after a report linked the company’s chatbot to a banned Chinese state-run telecommunications company. It’s not immediately clear whether the bill will be supported by other lawmakers. The Trump administration is also mulling a ban over national security concerns, a person familiar with the matter told Reuters. Some U.S. agencies have already restricted access to DeepSeek, including the U.S. Navy, the Defense Department, the Commerce Department and NASA, among others. The House’s Chief Administrative Officer reportedly claimed “threat actors are already exploiting DeepSeek to deliver malicious software and infect devices,” as some congressional offices have been warned not to use DeepSeek.

Key Background

DeepSeek released its advanced reasoning model R1 in January, a move that appeared to put China in close competition with U.S. tech and AI giants like OpenAI and Meta. R1 achieved performance on par with OpenAI’s o1 model across several benchmarks and reportedly exceeded o1’s in the MATH-500 test. DeepSeek’s R1 model is available for anyone to access, use, study, modify and share, as the open-source model surged to the top of the iPhone download charts—surpassing OpenAI’s ChatGPT. R1’s release sparked declines across several U.S. tech stocks, led by Nvidia closing down 17% on Jan. 27. Nvidia’s dive cut $589 billion off its market cap, marking the largest single-day loss of value for any public company in history. OpenAI CEO Sam Altman praised DeepSeek’s products and noted the R1 model was “impressive … particularly around what [DeepSeek is]able to deliver for the price.”