GPT-5 Reduces Cell-Free Protein Synthesis (CFPS) Costs

Unlike mathematics or physics, biology requires physical experiments, and the speed of experimental iteration creates a bottleneck for research. To address this, OpenAI connected GPT-5 to Ginkgo Bioworks' cloud lab, building an autonomous experimentation system that independently handles everything from experiment design to execution and learning.

This system was deployed to optimize the cell-free protein synthesis (CFPS) process. GPT-5 designed batches of experiments, robots carried them out, and the resulting data was fed back into the model to propose the next experiment—a closed-loop approach that repeated across a total of 6 experimental cycles.

In this process, the system tested more than 36,000 unique CFPS reaction compositions. As a result, GPT-5 achieved a new technical standard in just 3 experiments, cutting protein production costs by 40% and improving reagent costs by 57%.