Chinese AI company DeepSeek has released an updated version of its low-cost AI model.
The new model, named DeepSeek-V3-0324, is now available on the AI development platform Hugging Face, Reuters reported.
Go deeper with GlobalData
The company stated that DeepSeek-V3-0324 offers significant advancements over its predecessor with improved reasoning capabilities, stronger front-end development skills, and more sophisticated tool-use functionalities.
DeepSeek introduced its original V3 model in December 2024, followed by the R1 model in January 2025.
Both versions have been recognised for offering cost-effective alternatives to leading Western AI models.
The performance of DeepSeek’s models is claimed to rival that of advanced systems developed by US-based firms such as OpenAI and Meta Platforms.
According to DeepSeek, the R1 model, in particular, operates at a fraction of the cost of comparable Western offerings.
The company claims that R1 costs 20 to 50 times less than OpenAI’s o1 model, depending on the task. However, these cost and performance assertions have drawn scepticism from some observers in the global AI sector.
Meanwhile, DeepSeek has faced growing international scrutiny. Earlier in March 2025, reports of the US government considering measures to restrict the company’s applications on government-issued devices due to national security concerns surfaced.
The proposed rule is claimed to block the installation of DeepSeek’s chatbot app on official US devices.
This development comes in the wake of government officials being concerned about the company’s handling of user data, which the company stores on servers located in China.
Some officials have raised concerns about the company’s transparency in collecting and processing user data, as well as who has access to it.
In January 2025, DeepSeek was hit with a cyberattack putting its new user registrations for its V3 platform on halt.
