OpenAI’s COBOL migration score has made a striking leap
This week, OpenAI put COBOL in the spotlight. It announced benchmark results suggesting its latest models are getting much better The post OpenAI’s COBOL migration score has made a striking leap appeared first on The New Stack .
OpenAI has announced impressive advancements in its ability to migrate COBOL programs and generate software using less code and documentation. In a benchmark test, GPT-6.1 Sol achieved an 86.6% accuracy at low reasoning effort, a significant improvement from GPT-5.6 Sol's 5.4%. At higher reasoning effort, the newer model reached 97.9%, though the evaluation details remain undisclosed.
The company highlighted these results as substantial improvements, capable of delivering higher accuracy at lower cost, which could benefit teams maintaining and modernizing legacy software. However, the lack of public COBOL test cases means developers cannot independently verify these findings. Despite OpenAI's progress, COBOL remains a challenging target for AI coding tools due to limited code and documentation compared to more modern languages like Python and JavaScript.
Experienced COBOL developers are also scarce, and documentation is often outdated. Additionally, while GPT-6.1 Sol showed notable improvements across 20 programming languages, Anthropic, AWS, and Amazon have also been working on AI-assisted COBOL modernization efforts.
Written by urgent.news from The New Stack's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
