DeepSeek updates its R1 reasoning AI model, releases it on Hugging Face
TechCrunchMAY 22
DeepSeek Launches New R1 Model
Above: An illustration photo shows the DeepSeek LOGO displayed on a smartphone in Suqian City, Jiangsu Province, China, on May 29, 2025. Image copyright: CFOTO/Contributor/Future Publishing via Getty Images
The Spin
Anti-China narrative
While DeepSeek's rapid rise is commendable from a technological standpoint, the AI model is still only as good or worse than its counterparts. Its biggest flaw — biased and inaccurate information — is related to its main cybersecurity issue: the Chinese government controls its every move. There's no need to use DeepSeek when there are multiple alternatives that actually tell the truth.
Pro-China narrative
This is the pot calling the kettle black. While DeepSeek has single-handedly offered top-tier, cost-effective AI models to the world, American AI companies have raised prices and kept their source code secret to boost profits. China is the one democratizing AI, not the other way around. The West can sling as many accusations as it wants, but it won't change the fact that Chinese AI is rising to the top.