OpenAI says GPT-6.1 Sol is far better at migrating legacy code
Original title: OpenAI says GPT-6.1 Sol is dramatically better at migrating legacy code
OpenAI's own MultiLangBench shows GPT-6.1 Sol at 86.6% COBOL migration accuracy versus 5.4% for GPT-5.6 Sol.
OpenAI said newer models write and migrate code in older languages, including COBOL, Fortran, and Visual Basic, better than earlier models. On its MultiLangBench, which asks models to build a SQLite-compatible database engine across 20 languages, GPT-6 Astra averaged a 87.4% test pass rate, up from 26.8% for GPT-5.6 Sol. OpenAI also said GPT-6.1 Sol reached 86.6% accuracy migrating COBOL at low reasoning effort and 97.9% at higher effort, and that production migrations would still need expert review.
Unconfirmed: The pass rates and migration accuracy figures are OpenAI's own benchmark results, not independently verified.
- COBOL pass rate rose from 10.9% to 72.1%; Fortran from 21.4% to 83.5%.
- Visual Basic rose from 27.5% to 89.4%, and OCaml from 25.6% to 85.8%.
- OpenAI says COBOL still supports 95% of ATM transactions.
- Higher-effort COBOL migration accuracy reached 97.9%, near saturation.
- Vendors
- OpenAI
- Products
- MultiLangBench
- AI models
- GPT-6.1 SolGPT-6 AstraGPT-5.6 Sol
From the original article
1 min read at bleepingcomputer.com
OpenAI confirmed that its newer models are much better at writing and migrating software in older programming languages. [...]
By Mayank Parmar · bleepingcomputer.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.