DeepSeek: Everything you need to know about the AI chatbot app

So apparently, a Chinese hedge-fund-backed AI lab called DeepSeek has just shot to the top of both the App Store and Google Play charts with their chatbot—and it’s not just hype. Their latest model, DeepSeek V3, is outperforming big names like Llama and holding its own against GPT-4o on some benchmarks.

Here’s the wild part:

• They built their own data centers
• They’re running on downgraded Nvidia chips due to export bans
• They aren’t taking VC money
• And yet… they’re crushing it on price and performance

They’ve even released a “reasoning” model (R1) that fact-checks itself and is being used to build over 500 derivative models already.

Of course, it’s not all smooth sailing—some governments and companies are banning them, citing propaganda and data privacy concerns, and their models are subject to China’s content regulations. But still, this is probably the first time we’ve seen a serious non-U.S. challenger really shake the market. Nvidia’s stock dropped 18% in January partly due to DeepSeek, and even Altman commented.

Could this be the first real AI “space race”? Thoughts?