OpenAI Updates AI Model for Legacy Coding Tasks
The upgraded model assists enterprises with maintenance of outdated software systems like COBOL and Fortran.
Updated on Oct. 19, 2026 in Artificial Intelligence
Live Poll
Do you trust artificial intelligence to accurately modernize critical legacy software systems used by businesses?
OpenAI has updated its artificial intelligence model to improve programming performance in legacy languages including COBOL, Fortran, and Visual Basic. The update aims to help organizations manage maintenance costs and address the shrinking workforce of engineers skilled in older codebases.
Why it matters
Enterprises struggle with the rising costs of supporting legacy infrastructure that remains critical to business operations. By improving the ability of AI to translate and optimize this older code, the model seeks to preserve system behavior while modernizing internal workflows.
The updated model utilizes agentic AI workflows to assist teams in understanding and optimizing business logic written in COBOL, which currently sees Pass@1 research benchmarks peaking at 40%.
The players
OpenAI
An artificial intelligence research organization that develops large language models for commercial and industrial applications.
Anthropic
An AI research and safety company that competes in the generative model space and released legacy language updates earlier this year.
IBM
A global technology corporation that provides the watsonx Code Assistant tool specifically designed for translating legacy COBOL code into Java.
The details
The update focuses on preserving existing system behavior during software modernization, which is vital as organizations contend with the increasing scarcity of developers who understand older programming languages. OpenAI is currently gathering feedback from enterprise customers to determine which additional languages or workflows require future model improvements.
Timeline
February 2026: Anthropic released its own set of COBOL-related updates.
October 19, 2026: OpenAI officially announced the updated model for legacy code.
The Tech Race
This release intensifies the competition for enterprise software modernization, directly challenging established incumbents like the IBM watsonx Code Assistant. The industry is shifting from general coding support toward specialized capabilities that maintain critical legacy infrastructure.
Developers and IT managers can leverage these tools to reduce technical debt and simplify the maintenance of business-critical systems. Organizations may see potential workflow improvements, though they must carefully validate the AI-generated translations to ensure system stability.
The takeaway
Legacy languages like COBOL remain fundamental to global enterprise infrastructure despite their age. Companies should focus on using AI to bridge the knowledge gap created by the retiring workforce of expert programmers.
Further reading
For broader trends in enterprise model applications, explore the Artificial Intelligence section.
Live Poll
Do you trust artificial intelligence to accurately modernize critical legacy software systems used by businesses?
