3 min read
Meta has hired Trapit Bansal, a former OpenAI researcher, to join its newly formed AI superintelligence unit, according to a source familiar with the matter.
Bansal, who left OpenAI in June according to his LinkedIn profile, was instrumental in initiating reinforcement learning projects at the company alongside co-founder Ilya Sutskever. He played a foundational role in OpenAI’s early AI reasoning advancements and contributed significantly to the development of its first reasoning model, o1. His departure was confirmed by OpenAI spokesperson Kayla Wood.
Meta’s Growing Superintelligence Lineup
With Bansal joining, Meta is expected to push forward in building an advanced AI reasoning model. The goal is to compete with OpenAI’s o3 and DeepSeek’s R1.
Meta has not yet released a public reasoning model. It aims to catch up with, and possibly surpass, its rivals in this field. Bansal joins a strong team that includes former Scale AI CEO Alexandr Wang and ex-Google DeepMind researcher Jack Rae.
Others on the team include Nat Friedman, Daniel Gross, and OpenAI alumni Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai. Johan Schalkwyk, formerly of Sesame, has also joined.
Zuckerberg’s Bold AI Strategy
Meta CEO Mark Zuckerberg has been actively recruiting top-tier AI talent in recent months, reportedly offering compensation packages of up to $100 million to attract leading researchers. Although the financial details of Bansal’s deal are undisclosed, the high-stakes nature of these hires underscores Meta’s determination to become a leader in AI.
According to The Wall Street Journal and Bloomberg, Zuckerberg looked into buying several AI-focused startups. These included Safe Superintelligence, Thinking Machines Labs, and Perplexity.
However, none of the talks led to a final deal.
Meanwhile, OpenAI CEO Sam Altman said in a podcast that Meta has tried to recruit talent from his company.
He added, “none of our best people have decided to take him up on that.”
Why AI Reasoning Models Matter?
AI reasoning models are considered the next big step in artificial intelligence. They help systems “think” through complex problems before giving answers.
These models use advanced computing and require more processing time. This boosts accuracy, improves performance on benchmarks, and enhances real-world use.
OpenAI, Google, and DeepSeek have already made strong progress in this area.
Meta’s goal is to integrate such frontier models across its product suite, from virtual assistants to business AI tools. Clara Shih, formerly of Salesforce, is leading the company’s AI agent initiative, which will rely heavily on internal AI breakthroughs from this new team.
Meta is adding top researchers like Bansal to its AI team. The pressure is mounting to launch a public reasoning model that can compete with industry leaders.
At the same time, OpenAI is reportedly set to release an open AI reasoning model in the coming weeks. This could raise the competitive stakes even higher.
Meta aims to play a key role in the future of intelligent systems. But it’s still unclear whether it can match or outperform current frontrunners.
