DeepSeek, a Chinese AI startup, is making waves with its innovative approach to large-language models, claiming they rival offerings from tech titans like OpenAI and Meta—all while operating on a fraction of the budget and hardware. The company’s flagship R1 reasoning model promises performance “comparable” to OpenAI’s GPT-4, while its newly launched Janus Pro multimodal AI model reportedly outperforms competitors like Stable Diffusion and DALL-E 3.
DeepSeek’s ChatGPT alternative has already skyrocketed to the top of the App Store, shaking up the market. By January 27th, Nvidia’s shares had dropped 17 percent by mid-afternoon, underscoring the startup’s disruptive impact on financial markets. Powered by the state-of-the-art DeepSeek-V3 model, the AI assistant offers users capabilities like trip planning, text generation, and answering questions. However, the app’s surging popularity came with growing pains—DeepSeek has begun restricting new signups due to “malicious attacks.”
Founded in 2023 by Liang Wenfeng, DeepSeek has drawn global attention for its ability to build open-source AI models with less financial and computational investment. While industry giants like OpenAI and Google spend billions on their models, DeepSeek has managed to achieve similar results despite stringent U.S. export controls that bar companies like Nvidia from selling high-performance GPUs to China.
DeepSeek’s rise has thrown a spotlight on the potential of cost-efficient, open-source AI development, challenging the dominance of deep-pocketed tech behemoths. As the startup continues to scale—and fend off the growing pains of rapid success—the industry will be watching closely to see if DeepSeek’s models can live up to their promise and reshape the competitive landscape of artificial intelligence.