{"slug": "openai-upgrades-its-model-for-cobol-fortran-and-other-legacy-code", "title": "OpenAI upgrades its model for COBOL, Fortran, and other legacy code", "summary": "OpenAI said on October 19, 2026 that its newest model is better at writing and migrating code in COBOL, Fortran, and Visual Basic, targeting low-resource languages that still run banks, government agencies, and large enterprises. OpenAI framed the update around two jobs — helping teams understand decades-old business logic and preserving existing system behavior during modernization — and asked developers and enterprise customers which other languages or workflows should get similar improvements. OpenAI published no comparable benchmark figures with the announcement, and the update lands in a crowded field: Anthropic released COBOL-related updates in February 2026, and IBM shares fell approximately 13% after those updates, while an estimated 800 billion lines of COBOL remain in production.", "body_md": "OpenAI official logo (public domain, Wikimedia Commons) — CryptoBriefing brand treatment\n\n# OpenAI upgrades its model for COBOL, Fortran, and other legacy code\n\nThe update targets decades-old languages that still run banks, governments, and large enterprises\n\n[OpenAI](https://cryptobriefing.com/markets/openai/) says its latest model is now better at writing and migrating software in COBOL, Fortran, and Visual Basic.\n\nAn estimated 800 billion lines of COBOL remain in production. That code powers systems at banks, government agencies, and large enterprises.\n\n## What OpenAI announced\n\nOpenAI shared the update on X on October 19, 2026. The company said its newest model improvements help with building and migrating code in what it calls low-resource languages.\n\nIn AI terms, “low-resource” describes a language with relatively little code for a model to learn from.\n\nOpenAI framed the work around two jobs. The first is helping teams understand business logic that was written decades ago. The second is preserving how existing systems behave while they get modernized.\n\nOpenAI is also asking developers and enterprise customers for input. It wants to hear which other languages or workflows could benefit from similar improvements.\n\n## Why legacy code is so hard for AI\n\nCOBOL has historically been a weak spot for coding models. Research benchmarks have shown low Pass@1 and compilation rates on COBOL tasks. Those rates have ranged from single digits to about 40%.\n\n### AI, tech, and the markets they move—in one daily briefing.\n\nDaily. Free. Join 34,000+ readers across crypto, finance, and policy.\n\nPass@1 measures how often a model gets a working answer on its first try.\n\nLegacy systems often use domain-specific dialects. They also carry business rules that may not be documented anywhere outside the code itself.\n\nFortran shows where things could be heading. Agentic AI workflows built on OpenAI models can reportedly translate and optimize Fortran code on their own. The cost can be as low as a few USD per kernel.\n\n## A crowded race to modernize old systems\n\nOpenAI is not alone in chasing this market. [Anthropic](https://cryptobriefing.com/markets/anthropic/) released COBOL-related updates in February 2026. [IBM](https://cryptobriefing.com/markets/ibm/) offers watsonx Code Assistant, a tool built to translate COBOL into Java.\n\nIBM shares dropped by approximately 13% after Anthropic’s COBOL updates.\n\n## The workforce problem behind the push\n\nOrganizations running legacy systems face two pressures at once. Maintenance costs keep rising. The pool of engineers who know these languages keeps shrinking.\n\nSuccessful modernization usually takes both AI tools and human oversight. Verification is the critical piece.\n\n## What this means for enterprises and the tech sector\n\nIn regulated industries, a migration that changes behavior even slightly can trigger compliance headaches. Any AI tool in this space will be judged on how reliably it keeps outputs identical, not just on whether the new code compiles.\n\nOpenAI has not published comparable benchmark figures for its latest update in this announcement, so independent testing will matter.\n\n**Disclosure:** This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our\n\n[Editorial Policy](https://cryptobriefing.com/editorial-policy/).", "url": "https://wpnews.pro/news/openai-upgrades-its-model-for-cobol-fortran-and-other-legacy-code", "canonical_source": "https://cryptobriefing.com/openai-upgrades-model-cobol-fortran-legacy-code/", "published_at": "2026-10-19 22:34:19+00:00", "updated_at": "2026-10-19 22:46:58.491132+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-tools", "ai-products", "developer-tools"], "entities": ["OpenAI", "Anthropic", "IBM", "watsonx Code Assistant", "COBOL", "Fortran", "Visual Basic"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/openai-upgrades-its-model-for-cobol-fortran-and-other-legacy-code", "markdown": "https://wpnews.pro/news/openai-upgrades-its-model-for-cobol-fortran-and-other-legacy-code.md", "text": "https://wpnews.pro/news/openai-upgrades-its-model-for-cobol-fortran-and-other-legacy-code.txt", "jsonld": "https://wpnews.pro/news/openai-upgrades-its-model-for-cobol-fortran-and-other-legacy-code.jsonld"}}