Why OpenAI's reasoning chief has resigned
07 Jan 2026

Jerry Tworek, a leading researcher at OpenAI, has announced his resignation after nearly seven years with the company.

He made the announcement in a post on X, sharing an internal memo sent to his team.

The decision comes as Tworek wishes to explore research paths that are becoming increasingly difficult within OpenAI's current priorities.

Tworek's contributions to OpenAI's evolution
Impact

Joining OpenAI in 2019, when it was still largely a nonprofit research lab, Tworek became instrumental in the company's transition toward advanced reasoning models and practical AI systems for developers.

He led the development of the o1 reasoning model, which moved away from pure pretraining to reinforcement learning-based reasoning.

The approach proved successful as o1 topped LiveBench's leaderboard in August for complex reasoning tasks.

Tworek's role in OpenAI's developer ecosystem
Contributions

Apart from his work on reasoning models, Tworek also greatly contributed to OpenAI's developer ecosystem.

He led research on Codex, which became the backbone of GitHub Copilot, and directly contributed to the coding capabilities of GPT-4 and ChatGPT.

In his farewell note, he reflected on scaling reinforcement learning on robots and training early coding models as some of his key achievements at OpenAI.

Tworek's reflections on his time at OpenAI
Farewell

In his farewell note, Tworek expressed that his time at OpenAI was deeply formative, both personally and professionally.

He spoke about long nights, close friendships, and the unique opportunity to witness multiple foundational AI breakthroughs from within the organization.

While departures are never easy, he sees this moment as an opening for new ideas and renewed progress.