OpenAI built two evaluations. MultiLangBench asks a model to write a SQLite-compatible database engine from scratch in one of 20 languages, scored against tests adapted from SQLite's own suite.
In the News: October 20, 2026 (Extra 2)
OpenAI reports its newest models migrating COBOL programs with 86.6% accuracy on its own test, up from 5.4%, and releases a 20-language build benchmark.
Extra edition
Machine-readable
Download Markdown