United StatesHealthcare and life sciences2025 – ongoing

Ginkgo Bioworks

Autonomous cloud lab run by GPT-5 for cell-free protein synthesis

Ginkgo Bioworks and OpenAI cut the benchmark cost of cell-free protein by 40% with a GPT-5 lab

WorkingCompany's own claimOfficial ↗

Problem

Cell-free protein synthesis depends on reaction mixes with many components, and finding a cheaper mix takes many rounds of experiments. Ginkgo wanted to see whether an AI model could run that search with little human input.

Approach

Ginkgo and OpenAI connected GPT-5 to Ginkgo's cloud lab. The model designed experiments, the lab ran them and the model analyzed the results. Every proposed experiment was checked against a Pydantic model before it ran. Staff handled reagent preparation, loading and oversight. The work covered six cycles over about six months.

Results

$422 per gramcost of the benchmark protein sfGFP, against a previously reported $698 per gram (a 40% reduction), per GinkgoCompany's own claim[1]

36,000reaction compositions tested across six cycles, per GinkgoCompany's own claim[1]

more than 580384-well plates run, per GinkgoCompany's own claim[1]

Takeaway

The result comes from one benchmark protein, sfGFP, and a preprint that has not been peer reviewed. Ginkgo is selling the improved reaction mix.

Sources

Ginkgo Bioworks' Autonomous Laboratory Driven by OpenAI's GPT-5 Achieves 40% Improvement Over State-of-the-Art Scientific Benchmark
Ginkgo Bioworks (PR Newswire)5 Feb 2026Company publication

Source summaries

[1]Ginkgo Bioworks (PR Newswire)5 Feb 2026

Ginkgo's press release describes a GPT-5 driven autonomous lab that lowered the cost of cell-free production of sfGFP from $698 to $422 per gram. It reports 36,000 reaction compositions on more than 580 plates and nearly 150,000 data points. The findings are in a preprint that has not been peer reviewed.

Full details are in the original sources.

Added 2 Nov 2026 · Updated 2 Nov 2026