Meta has brought on Trapit Bansal, a prominent researcher from OpenAI, to join its new AI superintelligence division focused on developing reasoning models, according to a source speaking to TechCrunch.
OpenAI spokesperson Kayla Wood confirmed Bansal’s departure, and his LinkedIn profile indicates he left the company in June.
Trapit Bansal, who joined OpenAI in 2022, played a pivotal role in launching the company’s reinforcement learning efforts alongside co-founder Ilya Sutskever. He’s also credited as a core contributor to OpenAI’s first AI reasoning model, o1.
Bansal Joins Meta’s Elite AI Team to Boost Reasoning Model Efforts
His move to Meta is expected to be a major asset for the company’s new AI superintelligence unit, which already includes high-profile names like former Scale AI CEO Alexandr Wang. Meta is also reportedly in talks with ex-GitHub CEO Nat Friedman and Safe Superintelligence co-founder Daniel Gross. Bansal’s expertise could help Meta develop a next-generation reasoning model to rival top-tier systems like OpenAI’s o3 and DeepSeek’s R1. As of now, Meta hasn’t released a reasoning model of its own.
CEO Mark Zuckerberg has been aggressively expanding Meta’s AI team, reportedly offering compensation packages of up to $100 million to attract top talent. While it’s not known what Bansal was offered, his recruitment is part of a broader trend.
According to The Wall Street Journal, three other former OpenAI researchers—Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai—have also joined Meta’s AI superintelligence group. Bansal will be joining them, along with former Google DeepMind scientist Jack Rae and Johan Schalkwyk, previously a machine learning lead at the startup Sesame, as reported by Bloomberg.
Zuckerberg Explored Acquiring Top AI Startups to Bolster Superintelligence Unit
To expand its AI superintelligence division, Mark Zuckerberg reportedly pursued acquisitions of several high-profile AI startups, including Ilya Sutskever’s Safe Superintelligence, Mira Murati’s Thinking Machines Labs, and Perplexity. However, none of these discussions advanced to a finalized deal.
During a recent podcast appearance, OpenAI CEO Sam Altman remarked that Meta has made efforts to lure top talent away from his company, but stated that “none of our best people have decided to take him up on that.”
Meta declined to provide a comment on the matter.
AI Reasoning Models Emerge as a Top Priority for Meta’s Superintelligence Team
AI reasoning models are a critical focus for Meta’s new superintelligence team. Over the past year, companies like OpenAI, Google, and DeepSeek have released advanced models that demonstrate strong reasoning abilities, pushing the boundaries of what AI can achieve. These models improve performance by taking extra time and computing power to work through problems before delivering answers—an approach that’s shown success in both benchmark tests and real-world tasks.
Meta’s superintelligence lab has the potential to become a core driver of AI innovation across the company, similar to how DeepMind supports Google’s broader ecosystem. Meta is also pursuing a major initiative to build AI business agents, led by former Salesforce AI chief Clara Shih. To make these agents truly competitive, Meta must first develop state-of-the-art reasoning models to power them.
With hires like Bansal and other top AI experts, Meta aims to gain ground in the AI race. However, that goal may be challenged by OpenAI’s upcoming launch of an open AI reasoning model—an announcement that could raise the stakes for Meta’s own public AI tools.
Read the original article on: TechCrunch
Read more: Facebook Admins Report Widespread Bans, Meta Says It’s Working On a Fix