At 16, Mira Murati left Albania for an island off the coast of British Columbia. She had won a Davis United World Colleges scholarship to Pearson College, a school that gathers teenagers from around the world, and she went.
She was born in Vlore in December 1988, in the last years of the country's communist regime. What almost nobody pictures when they hear her name is how her house was run. Her parents were both high school literature teachers, and they encouraged her studies. Her own taste ran to maths and science. She has said she had an organic interest in them, and she competed in Olympiads through school.
Reports said Meta tried to hire away her researchers with packages of up to $1.5 billion; the same reports said none left.
After Pearson came Colby College, where she took a BA in mathematics in 2011, then a mechanical engineering degree at Dartmouth in 2012. A summer at Goldman Sachs in Tokyo, a spell at Zodiac Aerospace, and in 2013 a job as a product manager on Tesla's Model X. Leap Motion followed from 2016 to 2018. Then OpenAI, where by 2022 she was chief technology officer, overseeing ChatGPT, DALL-E, Codex and Sora.
In November 2023 she held the chief executive job for about three days after the board removed Sam Altman. Altman returned and she went back to her old role. In September 2024 she left, writing that there was never an ideal time to step away from a place one cherishes, yet this moment felt right.
She founded Thinking Machines Lab in February 2025. Within five months it had raised about $2 billion at a $12 billion valuation, and Albania's government added $10 million, which forced the country to amend its 2025 budget. Reports said Meta tried to hire away her researchers with packages of up to $1.5 billion; the same reports said none left.
Then they did. In January 2026, cofounders Barret Zoph and Luke Metz went back to OpenAI and Andrew Tulloch left for Meta. A reported $50 billion funding ask faded. In July 2026 the lab shipped Inkling, a 975-billion-parameter model under an Apache license, though it was partly built on Chinese open-weight models.
The sources are thin on how any of this felt, and it would be wrong to guess. What they do show is a girl from a family of literature teachers who kept choosing to leave a place she was good at. A founder can learn that the people you hire can leave in the same way, and the only lasting answer is the work they came to do.