{"slug": "openai-reports-benchmark-gains-for-cobol-and-other-legacy-language-coding", "title": "OpenAI reports benchmark gains for COBOL and other legacy-language coding", "summary": "OpenAI said in an October 19 post on X that its latest models, including GPT-6 Astra, have improved at writing and migrating code in COBOL, Fortran and Visual Basic, with GPT-6 Astra passing 72.1% of COBOL tests in OpenAI's SQLite-building evaluation. OpenAI described the work as a capability update rather than a standalone product, and its technical account says developers can try the models on coding tasks through OpenAI's API. OpenAI's separate migration results rely on OpenAI-designed hidden tests, leaving production safety and enterprise adoption of legacy-language migration unresolved.", "body_md": "# OpenAI reports benchmark gains for COBOL and other legacy-language coding\n\n**The company says GPT-6 Astra passed 72.1% of COBOL tests in its SQLite-building evaluation; its separate migration results use OpenAI-designed hidden tests.**\n\n        By [Ryan Merket](https://runtimewire.com/author/ryan-merket)\n        · Published \n\nPrimary source: [X](https://x.com/OpenAI/status/2112310345863930168)\n\n## Why it matters\n\nCOBOL migration tests whether coding models can preserve business behavior across costly modernization projects. OpenAI's results show progress on its own evaluations, while production safety and enterprise adoption remain unresolved.\n\nOpenAI says its latest models have improved at writing and migrating code in COBOL, Fortran and Visual Basic, targeting languages that remain embedded in financial and other business systems. In an [October 19th post on X](https://x.com/OpenAI/status/2112310345863930168), the company described the work as a capability update; it did not announce a standalone product. Its [technical account](https://openai.com/index/low-resource-programming-languages/) says developers can try the models on coding tasks through OpenAI's API.\n\nWhen you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.", "url": "https://wpnews.pro/news/openai-reports-benchmark-gains-for-cobol-and-other-legacy-language-coding", "canonical_source": "https://runtimewire.com/article/openai-low-resource-language-coding-benchmarks", "published_at": "2026-10-20 03:14:01+00:00", "updated_at": "2026-10-20 07:15:45.335448+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-research", "ai-tools", "generative-ai"], "entities": ["OpenAI", "GPT-6 Astra", "COBOL", "Fortran", "Visual Basic", "SQLite", "X"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/openai-reports-benchmark-gains-for-cobol-and-other-legacy-language-coding", "markdown": "https://wpnews.pro/news/openai-reports-benchmark-gains-for-cobol-and-other-legacy-language-coding.md", "text": "https://wpnews.pro/news/openai-reports-benchmark-gains-for-cobol-and-other-legacy-language-coding.txt", "jsonld": "https://wpnews.pro/news/openai-reports-benchmark-gains-for-cobol-and-other-legacy-language-coding.jsonld"}}