Meta CEO Mark Zuckerberg lauds DeepSeek, predicts breakthrough year for AI assistants
Meanwhile, Mark Zuckerberg said Meta's open-source platform Llama 4 is making great progress in training
ADVERTISEMENT
Meta Platforms Inc. CEO Mark Zuckerberg heaped praises on Chinese artificial intelligence startup DeepSeek during the Quarter 4 earning call held on Thursday.
Zuckerberg said, "I think there’s a number of novel things that they did that I think we’re still digesting. And there are a number of things that they have advances that we will hope to implement in our systems".
Zuckerberg further said in the AI segment every new company will have some new advances that the rest of the field will learn from. "And that’s sort of how the technology industry goes", he added.
"In AI, I expect this is going to be the year when a highly intelligent and personalized AI assistant reaches more than 1 billion people," he said.
Earlier, Sam Altamn, CEO of Microsoft-backed OpenAI also praised DeepSeek, calling it an "impressive model".
China's AI-model DeepSeek, which has operated on a shoestring budget, disrupted the stocks of US-based tech giants this week. It is reportedly as powerful as OpenAI's o1 model which showed impressive results in queries related to mathematics and coding.
DeepSeek's researchers claimed that it cost them only $6 million to train, a fraction of the over $100 million cost used by the OpenAI boss when discussing GPT-4.
Meanwhile, Zuckerberg said Meta's open-source platform Llama 4 is making great progress in training. "Llama 4 mini is done with pre-training and our reasoning models and larger model are looking good too. Our goal with Llama 3 was to make open source competitive with closed models, and our goal for Llama 4 was to lead. Llama 4 will be natively multimodal -- it's an omni-model -- and it will have agentic capabilities," he said.
Meta's Q4 revenue stood at $47.3 billion, up 21 percent year-on-year. The social media giant reported an ad revenue of $46.8 billion in the December quarter.
