OpenAI Says GPT-6 Astra Passes 87.4% of MultiLangBench Tests Across 20 Languages Including COBOL and Fortran
OpenAI reported on Oct. 19 that GPT-6 Astra averages 87.4% on MultiLangBench, a new GitHub benchmark that builds a SQLite-compatible engine across 20 low-resource languages, up from 26.8% for GPT-5.6 Sol. COBOL rises from 10.9 to 72.1 and Fortran from 21.4 to 83.5, OpenAI says. All figures are OpenAI's own.
Legacy-code migration is one of the few enterprise AI markets with clear budgets, and OpenAI is pitching straight at it — on a benchmark it built. In a Monday, October 19 post, OpenAI said GPT-6 Astra averages 87.4% across 20 low-resource languages on MultiLangBench, which asks models to build a SQLite-compatible database engine, up from 26.8% for GPT-5.6 Sol. COBOL rises from 10.9 to 72.1 and Fortran from 21.4 to 83.5, OpenAI reports. GPT-6 Luna averages 40.3% at an estimated $0.77 per sample against $3.48 for GPT-5.6 Sol.
On a COBOL-to-C++/Java/Kotlin/Python/Rust migration test scored on byte-for-byte output parity, GPT-6.1 Sol reaches 86.6% at low reasoning effort against 5.4% for GPT-5.6 Sol, at 61% of the API cost, and 97.9% at higher effort, OpenAI says. It released MultiLangBench on GitHub and says expert review is still needed. We covered Astra's launch separately.
Limits
- Every number is OpenAI's, on a benchmark it built.