DeepSeek, a Chinese startup, is making waves in the AI sector with its advanced large-language models that compete against major players like OpenAI and Meta. Their R1 reasoning model reportedly matches the performance of OpenAI’s models, while their Janus Pro multimodal AI model claims to outperform prominent tools such as Stable Diffusion and DALL-E 3. DeepSeek’s rise has even caused a notable drop in Nvidia’s shares. Founded by Liang Wenfeng in 2023, the startup emphasizes building open-source AI models with fewer financial and technical resources, all while navigating strict US export regulations on high-performance chips.
DeepSeek claims its R1 reasoning model and Janus Pro multimodal AI model deliver performance comparable to leading counterparts from OpenAI and Meta while being cost-efficient.
DeepSeek's rapid rise has had a knock-on effect on shares of established rivals, exemplified by Nvidia's stock dip, highlighting the startup's disruptive potential.
This Chinese startup has created notable open-source AI models and managed to achieve significant performance with lower spending on resources than industry giants.
The app's unexpected popularity has led to restrictions on signups, as DeepSeek seeks to combat malicious attacks amid a surge in downloads.
Collection
[
|
...
]