OpenAI says its model is better at COBOL, Fortran, and Visual Basic
OpenAI announced on X that its newest model includes improvements for writing and migrating code in COBOL, Fortran and Visual Basic. The company framed the work as helping teams understand decades‑old business logic and preserve system behavior while modernising legacy applications. OpenAI is soliciting feedback from developers and enterprise customers on other languages or workflows that could…
Key points
- OpenAI announced model improvements for COBOL, Fortran, and Visual Basic to aid legacy code understanding and migration.
- Estimated 800 billion lines of COBOL remain in production; past benchmarks showed Pass@1 rates from single digits to ~40%.
- Anthropic released COBOL updates in February 2026 and IBM offers watsonx Code Assistant; IBM shares dropped about 13% after Anthropic’s news.
The update targets an estimated 800 billion lines of COBOL still in production at banks, government agencies and large enterprises. Prior research benchmarks have shown Pass@1 rates for COBOL tasks ranging from single digits up to about 40%, indicating historically low success. OpenAI has not released comparable benchmark figures for the new update, so independent testing will be needed. Anthropic released COBOL‑related updates in February 2026, and IBM’s watsonx Code Assistant translates COBOL to Java; IBM’s shares fell roughly 13% after Anthropic’s announcement. Agentic AI workflows can reportedly translate and optimise Fortran code for a cost as low as a few USD per kernel.
OpenAI upgrades its model for COBOL, Fortran, and other legacy code
cryptobriefing.com · 19 October 2026
Loading the full article…
This text was published by cryptobriefing.com and written by Diego Almada Lopez. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- doubleAI releases debut code execution model and raises $220 million · 3 src
- Cloudflare launches Clairic extraction model on Workers AI preview · 2 src
- Fal launches FLUX 3 Max video model, 22× faster than Seedance 2.5 · 1 src
- DeepSeek’s Flash ties Anthropic’s Opus 5.5 in browser-agent benchmark · 2 src
- MiniMax releases M3.1 Flash model at $0.15 per 1M input tokens · 4 src
Comments
via GitHub Discussions