Thinking Machines Lab, launched in 2025 under the leadership of Mira Murati (former OpenAI CTO), focuses on advancing frontier AI systems with an emphasis on human-AI collaboration and multimodal capabilities. The organisation is built by a team of scientists and engineers whose prior work includes widely adopted AI products such as ChatGPT and Character.ai, as well as foundational open-source projects including PyTorch.
The lab's mission centres on three pillars: making frontier AI systems more widely understood, rendering them customizable to user needs and values, and enhancing their general capabilities. This orientation reflects a commitment to human-centred AI rather than systems designed for autonomous operation. The technical work spans artificial intelligence research, machine learning model training and interpretation, multimodal systems development, and open-source ML infrastructure.
Thinking Machines Lab has adopted an open science approach, releasing both research publications and code to broaden public understanding of how frontier systems are developed and trained. This stance distinguishes the organisation within a field often characterised by proprietary approaches, positioning it as committed to democratising access to knowledge and tools in AI development.
