OpenAI reports major gains on low-resource programming languages

OpenAI published results showing substantial improvements in coding and migration tasks for low-resource languages such as COBOL, Fortran, OCaml and Visual Basic. On MultiLangBench, GPT-6 Astra reached an average 87.4% test pass rate; on a COBOL migration evaluation, GPT-6.1 Sol achieved 86.6% accuracy at low reasoning effort. These languages underpin critical financial and scientific systems, so the measured gains expand practical utility for modernization workloads.