DeepSeek Just Did What OpenAI Wouldn’t

China’s DeepSeek has launched DeepSeek-R1, an MIT-licensed, fully open-source reasoning model that rivals OpenAI’s o1—and developers are losing it. Unlike OpenAI’s closed ecosystem, DeepSeek-R1 is free to modify, fine-tune, and commercialise, making it a dream come true for researchers and businesses alike.

“We are living in a timeline where a non-US company is keeping the original mission of OpenAI alive—truly open, frontier research that empowers all,” said Jim Fan, senior research manager and lead of embodied AI (GEAR Lab) at NVIDIA.

Besides the flagship model, DeepSeek also released six distilled versions, ranging from 1.5 billion to 70 billion parameters, optimised for math, coding, and reasoning. The lab is betting big on DeepSeek-R1-Zero, a model built entirely on reinforcement learning, which means it develops reasoning capabilities without any supervised data.

“Our goal is to explore the potential of LLMs to develop reasoning capabilities without any supervised data, focusing on their self-evolution through a pure RL process,” said the DeepSeek team.

And it’s working—DeepSeek-R1-Zero achieved a 71% pass rate on AIME 2024, up from just 15.6%, while DeepSeek-R1 surpassed OpenAI’s o1-1217 with a 79.8% Pass@1.

The AI community is taking notice. Paras Chopra, founder of Wingify, called it a breakthrough: “I love DeepSeek, so much! o1 level model is now open-source (MIT license).” Bindu Reddy, founder of Abacus AI, added: “DeepSeek R1 is on par with o1 and is open-source!! It blows my mind that Chinese make great, open and transparent tech.”

While OpenAI is caught up in o3 benchmark controversies, DeepSeek is setting new standards for transparency. “Whale ?? folks, respect,” said KissanAI founder Pratik Desai, summing up the industry’s reaction.

Enjoy the full story here.?

Nandan Nilakani is ‘Wrong’

AI wrapper Perplexity AI’s CEO Aravind Srinivas recently said that Infosys co-founder Nandan Nilekani is wrong in pushing Indians to ignore model training skills and focus on building on top of existing models. “Essential to do both,” Srinivas argued in a post on X.

“To be clear: Nandan Nilekani is awesome, and he’s done far more for India than any of us can imagine through Infosys, UPI, etc. But he’s wrong on pushing Indians to ignore model training skills…,” he said, countering Nilekani’s vision of making India the “AI use case capital” rather than investing in foundational model training.

The debate highlights a critical AI strategy dilemma: Should India focus on leading in research and foundational AI models or double down on application-driven growth? Read what experts have to say here.?

AI Agents Are Taking Over—Are You Ready?

SmartQ’s webinar ‘Building Enterprise Software Solutions at Lightning Speed with AI Agents’ is just around the corner, which will dive into how AI agents can increase developer productivity by 10x and transform enterprise software. With experts like Abhishek Ashok and Keshav Meda leading the session, this is your chance to stay ahead in the year of agentic AI. Don’t miss out—register now!

NVIDIA is an ‘Everything’ Company?

NVIDIA is no longer just about semiconductors—it’s into everything these days: healthcare, autonomous vehicles, AI computing, and now, its latest obsession—robotics.

“NVIDIA will be a robotics company at the end of the day, not just semiconductor,” said Dylan Patel, founder of SemiAnalysis. With Cosmos, a platform trained on 20 million hours of video, and its integration with Omniverse, NVIDIA is making AI that understands physics, movement, and real-world interactions—a foundation for the future of humanoid robots.?

From Jetson Thor for robotics computing to partnerships with Boston Dynamics, Figure AI, and XPENG Robotics, NVIDIA isn’t just building AI; it’s building the machines that will run on it.