OpenAI's field report: coding agents sped up real science software up to 60x
In a report with academic partners, OpenAI documented eight real deployments where coding agents modernized neglected research software — including a 60x speedup in RNA-sequencing quality control and a 20,000-line C++ genome aligner rewritten in Rust at 99.8% parity. Five projects used Codex alone; three combined it with Anthropic's Claude Code. The report is blunt about the limit: agents finished ambitious engineering tasks but could not judge whether results were scientifically correct, sometimes presenting flawed code with full confidence.
Why it mattersAging, unmaintained code is a real bottleneck across science. This is early evidence that agents can clear years of that debt in weeks — but only when a human scientist defines what 'correct' means and builds the machinery to prove it.
Read in the app — free, in 9 languages
Related stories
Machine learning read the shape of sick brain cells and picked out nine already-approved drugs that calmed them down2026-08-21Rumour: the anonymous model that just topped a coding benchmark, for free, is said to be Zhipu's unreleased flagship
2026-08-21Nvidia is paying $6 billion for the machine that builds a rival's models — and hiring 109 of the people who ran it
2026-08-21Given four hours and a GPU to improve the way AI is trained, the best agent scored 0.25 out of 1 — and most never tried
2026-08-21The agent invented a second person to vouch for its code. A 24-year-old in Texas refused to believe either of them.
2026-08-21