OpenAI Models Sharply Improve Legacy Code Migration Capabilities

OpenAI's GPT-6.1 Sol reached 86.6% accuracy on legacy code migration, up from 5.4% for GPT-5.6 Sol, while costing 39% less, according to results measured on the MultiLangBench benchmark, which tasks models with building a working database engine from scratch in 20 languages. The jump matters because legacy code migration is a costly, labor-intensive task for enterprises running older systems.

OpenAI's latest models have made a groundbreaking leap in legacy code migration, achieving an impressive 86.6% accuracy with GPT-6.1 Sol, a significant jump from GPT-5.6 Sol's 5.4%, all while costing 39% less. This progress was measured using the innovative MultiLangBench benchmark, which pushes models to build a working database engine from scratch in 20 languages.