Meta is building a team of AI experts — and they just scored a big win.
Trapit Bansal, a respected AI researcher from OpenAI, has joined Meta to help develop the company’s next big thing: AI models that can reason like humans.
He’ll be working with Meta’s new AI superintelligence unit, a team that’s quietly attracting some of the sharpest minds in artificial intelligence.
Who Is Trapit Bansal?
Bansal isn’t just any researcher.
He joined OpenAI in 2022 and played a big role in building the company’s early reinforcement learning systems. More importantly, he was one of the key contributors behind OpenAI’s first reasoning model, known as o1.
In the AI world, that’s a big deal.
This model laid the groundwork for newer, more powerful systems — like OpenAI’s o3 and DeepSeek’s R1 — which don’t just answer questions, but actually think through them first.
What’s Meta Planning?
Bansal is joining Meta’s AI superintelligence lab, a new division focused on building AI that’s more capable, more thoughtful, and more useful across Meta’s products.
He’ll work alongside some other heavy-hitters in the field, including:
- Former Scale AI CEO Alexandr Wang
- Machine learning experts like Lucas Beyer, Xiaohua Zhai, and Alexander Kolesnikov, who recently left OpenAI
- DeepMind’s former researcher Jack Rae
- AI veteran Johan Schalkwyk, who led ML at a startup called Sesame
Meta’s clearly not messing around. They’re building a team designed to compete with — or even surpass — today’s best AI labs.
Why Reasoning Models Matter
Most AI tools today are fast and responsive. But they’re not always good at thinking.
That’s where reasoning models come in. Instead of spitting out answers immediately, these models take a little extra time and computing power to:
- Break down the question
- Consider different possibilities
- Deliver a more thoughtful, accurate response
This is especially useful in areas like coding, science, education, and high-stakes decision-making — exactly where companies like OpenAI, Google DeepMind, and now Meta are focusing.
Meta’s Bigger AI Push
This hire is part of Mark Zuckerberg’s larger strategy to make Meta a top AI company.
He’s been offering massive compensation packages — sometimes worth $100 million or more — to lure top AI talent away from competitors.
Reports say he even tried to acquire AI startups like:
- Safe Superintelligence (co-founded by OpenAI’s Ilya Sutskever)
- Thinking Machines Labs (run by Mira Murati)
- Perplexity (an AI search company)
Those talks didn’t go anywhere, but it shows just how aggressive Meta is about building its AI team in-house.
OpenAI’s Response?
In a recent podcast, OpenAI CEO Sam Altman acknowledged Meta’s recruiting attempts.
He claimed that none of OpenAI’s best researchers had left — though recent reports suggest otherwise.
With people like Bansal, Beyer, and Zhai now at Meta, it’s clear the competition for AI talent is heating up.
What This Means Going Forward
Meta doesn’t yet offer its own AI reasoning model to the public, but that may change soon.
With Bansal and others on board, it’s likely we’ll see Meta:
- Launch a powerful reasoning model to compete with OpenAI and Google
- Use this technology to improve AI assistants, business tools, and smart agents
- Build out a DeepMind-style research group that influences all of Meta’s platforms
And with OpenAI reportedly preparing to release a new open-source reasoning model soon, Meta may be racing the clock to keep up.
Quick Recap
- Meta has hired Trapit Bansal, a top OpenAI researcher known for building early AI reasoning models.
- He joins a growing team of AI experts at Meta, including other former OpenAI and DeepMind staff.
- Meta is investing heavily in AI reasoning models, which help machines think more logically and make better decisions.
- This move puts Meta in direct competition with OpenAI, DeepMind, and DeepSeek.
- The AI talent war is in full swing — and Meta is clearly aiming for the top