Skip to content
NewA calmer, faster ZeroHour

OpenAI says newer models preserve behavior in COBOL migration tests

Original title: Can AI help solve the COBOL code crisis?

The Verge · AIby Robert Hart ()6 min read1
AI summary · x-ai/grok-4.7Summaries by Vani AI

GPT-6.1 Sol hit 98% behavior preservation at higher reasoning; researchers say that is far from replacing live systems.

On October 19, OpenAI said newer models can migrate COBOL and other legacy languages to C++, Java, Kotlin, Python, and Rust without changing what the software does. In its tests, GPT-6.1 Sol reached 87% behavior preservation at low reasoning and 98% at higher effort, while GPT-6 Sol scored 21% and GPT-5.6 Sol 5% at low effort. Anthropic has pitched Claude Code for COBOL modernization without publishing benchmarks. Researchers told The Verge that rewrite scores do not show critical systems can be safely replaced.

Unconfirmed: Accuracy figures are OpenAI’s own tests; researchers called the tasks unlike real legacy migrations.

Key facts
  • IBM says COBOL supports 95% of ATM transactions and 80% of in-person card transactions.
  • OpenAI says a migration counts as successful only if all tests pass.
  • Anthropic published a code-modernization playbook in February without performance numbers.
  • IBM shares fell after Anthropic’s announcement, The Verge reports.
  • An OpenAI spokesperson said legacy code is still far from a solved problem.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

6 min read at theverge.com

A computer running COBOL 73, an antiquated computer programming language, awaits taxpayer information entries at the Internal Revenue Service in Austin, Tx. , on Friday, June 10, 2022. | Image: Getty Images What do the IRS, banks, airlines, and government infrastructure around the world have in common?

The Verge · AI

By Robert Hart · theverge.com

Read the full article at The Verge · AI

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.