Interpret literature and plan a protocol
Information tasks have benchmark coverage; physical success requires a different test.
SCIENCE UNDER AUTOMATION
An agent can reason about a protocol while an experiment still depends on instruments, specimens and troubleshooting. Automation becomes more plausible where those physical steps are standardized and connected.
40% lower protein-production cost reported
One protein (sfGFP), one CFPS system. Thousands of compositions are not thousands of independent protein workflows.
Iterative CFPS optimization achieved measured cost improvements.
GPT-5 / Ginkgo cell-free protein synthesisCloud-lab execution relies on a prepared platform and lab support.
GPT-5 / Ginkgo cell-free protein synthesisRepeated measured feedback supports this specific assay.
GPT-5 / Ginkgo cell-free protein synthesisLaboratory reasoning tasks support planning, not physical execution.
LABBench2 v2General physical autonomy is not evaluated by LABBench2.
LABBench2 v2The cited evaluations do not establish autonomous mechanism validation.
LABBench2 v2Information tasks have benchmark coverage; physical success requires a different test.
Requires instrument access and evidence from the actual experimental setting.
A convincing account depends on discriminating tests and independent validation.
The results above measure specific tasks. The field assessment also requires completion of the connected workflow, including the requirements listed here. A field-wide replacement percentage has not been established.
Read the assessment method