According to OpenAI, GPT-5 was linked to an autonomous lab in a closed-loop setup where it designed experiment batches, the lab executed them, and the results informed subsequent designs across six iterations. OpenAI reports the system explored over 36000 reaction compositions across 580 automated plates, reflecting iterative design-execute-learn workflows at laboratory scale.

According to OpenAI, GPT-5 was connected to Ginkgo’s autonomous lab to propose experiments, run them at scale, learn from outcomes, and decide subsequent iterations in a closed loop, resulting in a reported 40% reduction in protein production costs (source: OpenAI). For traders, the claim highlights accelerating AI-driven lab automation and synthetic biology efficiency that could influence biomanufacturing economics and demand for AI infrastructure and lab automation software if reproduced at scale (underlying facts source: OpenAI).