{"schemaVersion":"1.0","generatedFrom":"https://brightaifuture.com/discoveries/openai-cobol-ai-legacy-software-multilangbench","record":{"id":"openai-cobol-ai-legacy-software-multilangbench","headline":"OpenAI’s COBOL AI gains put legacy software in focus","canonicalUrl":"https://brightaifuture.com/discoveries/openai-cobol-ai-legacy-software-multilangbench","datePublished":"2026-10-19","dateModified":null,"sourcePublicationDate":"2026-10-19","author":null,"publisher":{"name":"Bright AI Future","url":"https://brightaifuture.com/"},"topics":[],"summary":"New coding results invite a serious look at AI-assisted maintenance. Bright examines the evidence, the limits and what safer modernization would require.","evidenceState":"Demonstrated","keyFacts":[{"label":"AI’s role","value":"OpenAI evaluates code translation and SQLite-compatible database implementations across programming languages."},{"label":"Documented result","value":"OpenAI reports 86.6% COBOL migration accuracy for GPT-6.1 Sol at low reasoning effort and 97.9% at higher effort; GPT-6 Astra averaged an 87.4% MultiLangBench test pass rate at maximum effort across 20 languages. The measurements use different denominators."},{"label":"Important limitation","value":"Company-reported benchmark results; not independently reproduced by Bright."}],"limitations":["Company-reported benchmark results; not independently reproduced by Bright.","Benchmark scores do not establish the success probability of a live migration, production assurance, jobs eliminated or specialist time saved.","COBOL migration accuracy and MultiLangBench test pass rates use different denominators.","Teams must disclose test subsets and evaluate platform, transaction, integration and failure behavior for their own deployments."],"evidenceLinks":[{"title":"OpenAI research: low-resource programming languages","url":"https://openai.com/index/low-resource-programming-languages/","type":"institution"},{"title":"GAO legacy systems review","url":"https://www.gao.gov/products/gao-25-107795","type":"government"},{"title":"UFS Weather Model documentation","url":"https://ufs-weather-model.readthedocs.io/en/develop/InputsOutputs.html","type":"institution"},{"title":"UFS Weather Model repository","url":"https://github.com/ufs-community/ufs-weather-model","type":"institution"},{"title":"OpenAI MultiLangBench repository","url":"https://github.com/openai/MultiLangBench","type":"institution"},{"title":"How SQLite is tested","url":"https://www.sqlite.org/testing.html","type":"institution"},{"title":"IBM COBOL overview","url":"https://www.ibm.com/think/topics/cobol","type":"institution"}],"evidencePackUrl":"https://brightaifuture.com/evidence-pack/openai-cobol-ai-legacy-software-multilangbench","embedUrl":"https://brightaifuture.com/embed/story/openai-cobol-ai-legacy-software-multilangbench","attribution":{"credit":"Bright AI Future","requirements":["Link to the canonical Bright record.","Keep material limitations with the claim they qualify.","Link to the original evidence when repeating a substantive claim.","Do not describe a source check or organization-reported result as independent verification."],"sourceRights":"Linked source material, quotations, trademarks and media remain subject to their owners’ terms. No reuse right is granted for third-party media."}},"claim":{"humanProblem":"Organizations depend on established software that must remain reliable while engineers understand, maintain and modernize it.","priorConstraint":"Established languages can have comparatively little model training material, while modernization also requires domain expertise and operational assurance.","aiRole":"OpenAI evaluates code translation and SQLite-compatible database implementations across programming languages.","documentedResult":"OpenAI reports 86.6% COBOL migration accuracy for GPT-6.1 Sol at low reasoning effort and 97.9% at higher effort; GPT-6 Astra averaged an 87.4% MultiLangBench test pass rate at maximum effort across 20 languages. The measurements use different denominators.","whyItMayMatter":"Better assistance could help maintainers understand, test or repair important software; dependable outcomes still require human engineering decisions.","unresolvedQuestions":["Can teams demonstrate dependable deployments and a reduction in total workload including review and repair?"]},"evidenceAssessment":{"state":"Demonstrated","claimConfidence":"unassessed","reviewState":"approved","reviewMethod":"ai-assisted","reviewNote":"Owner authorized separate publication of complete reviewed article. Company-reported benchmark results; not independently reproduced by Bright. Source review is recorded in the supplied approved article. No benchmark was run. Requested SEO title: OpenAI COBOL AI Results: What Legacy Software Teams Should Know. Requested meta description: OpenAI’s COBOL and MultiLangBench results raise hopes for legacy software. Bright explains the evidence, limits and practical steps for engineering teams.","lastSourceReview":"2026-10-19","independentVerification":"not-established-by-this-source-review"},"sources":[{"id":"source:openai-cobol-ai-legacy-software-multilangbench:openai","title":"OpenAI research: low-resource programming languages","type":"institution","url":"https://openai.com/index/low-resource-programming-languages/"},{"id":"source:openai-cobol-ai-legacy-software-multilangbench:gao","title":"GAO legacy systems review","type":"government","url":"https://www.gao.gov/products/gao-25-107795"},{"id":"source:openai-cobol-ai-legacy-software-multilangbench:ufs-docs","title":"UFS Weather Model documentation","type":"institution","url":"https://ufs-weather-model.readthedocs.io/en/develop/InputsOutputs.html"},{"id":"source:openai-cobol-ai-legacy-software-multilangbench:ufs-repo","title":"UFS Weather Model repository","type":"institution","url":"https://github.com/ufs-community/ufs-weather-model"},{"id":"source:openai-cobol-ai-legacy-software-multilangbench:benchmark","title":"OpenAI MultiLangBench repository","type":"institution","url":"https://github.com/openai/MultiLangBench"},{"id":"source:openai-cobol-ai-legacy-software-multilangbench:sqlite","title":"How SQLite is tested","type":"institution","url":"https://www.sqlite.org/testing.html"},{"id":"source:openai-cobol-ai-legacy-software-multilangbench:ibm","title":"IBM COBOL overview","type":"institution","url":"https://www.ibm.com/think/topics/cobol"}],"revisions":[{"id":"openai-cobol-ai-legacy-software-multilangbench-initial","recordedAt":"2026-10-19T23:01:45.983Z","sourceIds":["source:openai-cobol-ai-legacy-software-multilangbench:openai","source:openai-cobol-ai-legacy-software-multilangbench:gao","source:openai-cobol-ai-legacy-software-multilangbench:ufs-docs","source:openai-cobol-ai-legacy-software-multilangbench:ufs-repo","source:openai-cobol-ai-legacy-software-multilangbench:benchmark","source:openai-cobol-ai-legacy-software-multilangbench:sqlite","source:openai-cobol-ai-legacy-software-multilangbench:ibm"],"summary":"Initial owner-authorized reviewed article on OpenAI COBOL and MultiLangBench results, legacy-software context, benchmark limitations and practical engineering evaluation."}],"corrections":[]}