DeepSeek, a rising star in the tech world from China, has grabbed everyone’s attention with its remarkable R1 model. This model stands out by surpassing OpenAI’s o1 reasoning model in domains like math, science, and coding, all while operating at just 3% of the usual cost.
What sets DeepSeek apart is its approach to AI development. Unlike giants like Microsoft and OpenAI, which demand significant resources and intensive training, DeepSeek offers an open-source model. This makes it accessible for those who want to study and duplicate its achievements without breaking the bank. Its performance has been benchmarked against big names like GPT-4o, Gemini, and Claude, earning applause for its high efficiency.
Naturally, these developments have sparked excitement across the AI field, leaving major industry players both impressed and intrigued by its ability to combine performance with affordability. Yann LeCun, Meta’s AI lead scientist, shared his thoughts through Business Insider, appreciating DeepSeek’s contribution:
“They innovated by building on existing research. Since their work is open-source, it benefits everyone. That’s the strength of open research.”
The impact of DeepSeek’s advances has been significant enough to shake confidence in US supremacy in the AI sector. This turbulence hit shortly after OpenAI and SoftBank declared their intentions to invest a staggering $500 billion into the Stargate project, aiming to enhance AI infrastructure in the US.
LeCun highlights that DeepSeek’s success extends beyond merely outpacing the US in AI innovation. It underlines the continuing value of open-source AI models, hinting at a revival of OpenAI’s original mission focused on openness. He went on to stress that open-source models often outperform their proprietary counterparts.
R1 from DeepSeek is among a small group of open-sourced models, including Meta’s Llama. It’s interesting to note that OpenAI’s CEO, Sam Altman, hinted at plans to offer open-source products too—notably ironic, given that OpenAI initially championed open-source principles but has shifted towards a closed approach, underscoring safety as a key reason for this transition.
Recently, DeepSeek released its app, which swiftly captured the interest of many users. It even managed to outshine the wildly popular ChatGPT, claiming the title of the top free app download in the US. Meanwhile, major tech companies like Nvidia, Microsoft, and Meta have felt the tremors of DeepSeek’s ascent with notable drops in their stock prices, reflecting the disruptive chatter within the AI community.