# OpenAI details coding gains in low-resource languages and open-sources MultiLangBench



OpenAI



In an article dated October 19, 2026, OpenAI released the MultiLangBench evaluation setup and test suite on GitHub. The benchmark tests whether models can build a SQLite-compatible database engine across 20 programming languages. At maximum reasoning effort, GPT-6 Astra averaged an 87.4% test pass rate, compared with 26.8% for GPT-5.6 Sol. In a separate COBOL migration evaluation, GPT-6.1 Sol reached 97.9% accuracy at higher reasoning effort. These are OpenAI’s evaluation results, not a new model launch; production migrations still require expert review and testing.

- Review updated: Oct 20, 2026 08:39 Tue · Beijing Time (UTC+8)

## Sources

- Source excerpt: https://openai.com/index/low-resource-programming-languages
- Source excerpt: https://x.com/OpenAI/status/2112310345863930168

Canonical entry: https://www.ailyre.com/source/?id=openai-low-resource-programming-languages-2026-10-19&lang=en