--- title: OpenAI says its latest models are better at COBOL,... description: OpenAI says recent model improvements boost performance on low-resource programming languages like COBOL, Fortran, and Visual Basic, which still power core... canonical: https://daily.dev/posts/openai-says-its-latest-models-are-better-at-cobol-fortran-and-visual-basic-ls8tgwqqk twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: OpenAI says its latest models are better at COBOL, Fortran and Visual Basic | daily.dev og:description: OpenAI says recent model improvements boost performance on low-resource programming languages like COBOL, Fortran, and Visual Basic, which still power core... og:url: https://daily.dev/posts/openai-says-its-latest-models-are-better-at-cobol-fortran-and-visual-basic-ls8tgwqqk og:image: https://api.daily.dev/og/posts/lS8TGWQqk.png og:image:alt: OpenAI says its latest models are better at COBOL, Fortran and Visual Basic og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # OpenAI says its latest models are better at COBOL, Fortran and Visual Basic **[Collections](https://daily.dev/sources/collections)** · 1 min read · 0 upvotes · 0 comments ## Summary OpenAI says recent model improvements boost performance on low-resource programming languages like COBOL, Fortran, and Visual Basic, which still power core systems at banks, governments, and businesses. The company frames this as helping developers modernize legacy software while preserving existing business logic, and says it is soliciting input from teams working on these systems about which languages and workflows to prioritize next. No specific benchmarks or models were named in the announcement. ## Content OpenAI says its latest model advances improve how well its models build and migrate software in low-resource programming languages, naming COBOL, Fortran, and Visual Basic. These languages run software that banks, governments, and businesses still depend on. OpenAI says progress here means understanding decades of accumulated business logic and helping developers modernize those systems while preserving what already works. The company says it is shaping these capabilities around problems developers and enterprise customers bring to it. It is also asking teams who work on these systems to tell it which languages and workflows to take on next. The announcement was reposted by @charliermarsh. OpenAI hasn't named specific benchmarks or models in the announcement. Migration is the hard part of legacy work, though, so I'm curious what "preserving what works" looks like in practice. ## Questions this post answers ### Can OpenAI's latest models help migrate COBOL code to modern languages? OpenAI states its latest model advances improve performance on low-resource programming languages including COBOL, Fortran, and Visual Basic, aiming to help developers understand decades of accumulated business logic in legacy systems and modernize them while preserving existing functionality. No specific benchmark results or model names were disclosed to support this claim. _Teams weighing AI tools for legacy migrations can track real-world results on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 2 discussions and 447 comments across x (as of 2026-10-20). **TL;DR:** Replies mix jokes about COBOL bankers and retiring mainframe veterans with substantive skepticism that the real challenge is undocumented business logic and verifying behavioral equivalence, not language syntax; a vocal subset is also just annoyed OpenAI didn't ship a flashier new model instead. **Sentiment:** 25% positive · 40% mixed · 35% skeptical **The case for** - Several see real value in using AI to document and recover undocumented business logic buried in old systems before touching the code. - Some think AI-generated regression/behavioral test suites around legacy code could be more valuable than the migration itself. - A few frame this as a practical, high-value use case compared to flashy demos. **The pushback** - Many argue the hard part was never the syntax but 40 years of undocumented business rules and edge cases that no model can just infer. - Commenters question how equivalence is actually verified, e.g. packed-decimal rounding, numerical convergence, and matching production outputs exactly. - Others note real migrations also depend on surrounding infrastructure like JCL, copybooks, and batch job scheduling that go beyond just reading COBOL. - Some are skeptical of probabilistic models being trusted with high-stakes banking logic without formal verification. - A number of replies are just off-topic complaints that OpenAI should instead fix or release better flagship models. **By community** - x (mixed): Discussion splits between jokes/skepticism about banks and legacy coders, genuine technical concerns about verifying migrated business logic, and unrelated complaints about other OpenAI models. **Hottest debate:** Whether the real bottleneck in legacy migration is model capability at all, versus the undocumented business logic and verification/testing problem that no language model can simply solve. **Open questions** - How exactly is business-logic preservation verified during migration (e.g. packed-decimal rounding, numerical convergence, edge cases)? - Will the surrounding infrastructure (JCL, copybooks, batch scheduling, PLC ladder logic) be addressed alongside core language translation? - Which specific benchmarks or eval methods were used to produce the cited migration-accuracy improvements? **Highlights** > @OpenAI COBOL or any other of those languages have never been the issue, it's 40 years of undocumented business rules, and the one guy who knew why a field goes negative some days retired in 2009 LOL > — [pggalaviz on x · 33 points, 1 comments](https://x.com/pggalaviz/status/2112322862606381557) > @OpenAI COBOL migration is a custody problem first. Those systems hold 40 years of business logic nobody fully documented. An agent that moves code it cannot explain creates the same risk as a wire with no audit trail: it works until an examiner asks why. > — [realNickPawlak on x · 1 points](https://x.com/realNickPawlak/status/2112331442613358840) > @DaiLin67911 @OpenAI Fair question. Run old and new on the same day's transactions and diff every output. Each mismatch is an unwritten rule. > — [JoinPond on x · 1 points, 1 comments](https://x.com/JoinPond/status/2112331638558400746) > @OpenAI Translating legacy COBOL and Fortran is a formal verification problem. Probabilistic models guessing 40-year-old banking edge cases without formal proof engines and deterministic execution constraints are a recipe for silent operational failure. > — [RisingAIGlobal on x · 1 points](https://x.com/RisingAIGlobal/status/2112316115791856016) > @OpenAI how are you testing that business logic is actually preserved? matching outputs on old test data only catches the cases someone thought to test > — [ItsPopyn on x · 1 points](https://x.com/ItsPopyn/status/2112310595215364242) **Source threads** - [x](https://x.com/charliermarsh/status/2112314267890315462) · 0 points · 0 comments - [x](https://x.com/OpenAI/status/2112310345863930168) · 2 points · 447 comments ## Similar posts on daily.dev - [How AI helps break the cost barrier to COBOL modernization](https://daily.dev/posts/how-ai-helps-break-the-cost-barrier-to-cobol-modernization-h7oqac1ou) · Claude · 2 upvotes · 0 comments - [The Programming Languages AI Still Gets Surprisingly Wrong](https://daily.dev/posts/the-programming-languages-ai-still-gets-surprisingly-wrong-lbxoqyzy8) · Code Like A Girl · 4 upvotes · 1 comments --- Tags: [#openai](https://daily.dev/tags/openai), [#fortran](https://daily.dev/tags/fortran) [View this post on daily.dev](https://daily.dev/posts/openai-says-its-latest-models-are-better-at-cobol-fortran-and-visual-basic-ls8tgwqqk) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"OpenAI says its latest models are better at COBOL, Fortran and Visual Basic","url":"https://daily.dev/posts/openai-says-its-latest-models-are-better-at-cobol-fortran-and-visual-basic-ls8tgwqqk","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/openai-says-its-latest-models-are-better-at-cobol-fortran-and-visual-basic-ls8tgwqqk"},"datePublished":"2026-10-19T22:46:03.566Z","dateModified":"2026-10-20T03:17:17.575Z","description":"OpenAI says recent model improvements boost performance on low-resource programming languages like COBOL, Fortran, and Visual Basic, which still power core...","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/openai-says-its-latest-models-are-better-at-cobol-fortran-and-visual-basic-ls8tgwqqk","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"openai,fortran","timeRequired":"PT1M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"OpenAI says its latest models are better at COBOL, Fortran and Visual Basic"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/openai-says-its-latest-models-are-better-at-cobol-fortran-and-visual-basic-ls8tgwqqk#faq","mainEntity":[{"@type":"Question","name":"Can OpenAI's latest models help migrate COBOL code to modern languages?","acceptedAnswer":{"@type":"Answer","text":"OpenAI states its latest model advances improve performance on low-resource programming languages including COBOL, Fortran, and Visual Basic, aiming to help developers understand decades of accumulated business logic in legacy systems and modernize them while preserving existing functionality. No specific benchmark results or model names were disclosed to support this claim. Teams weighing AI tools for legacy migrations can track real-world results on daily.dev."}}]} ```