OpenAI says its models improved at migrating old code
OpenAI reported large gains in coding and legacy software migration, with one model scoring 97.9% on a COBOL test. Its work covers programming languages with fewer resources.
By Standard Topic. AI-generated from the reporting listed under Coverage. Published . Updated .
This story is no longer being updated. It reflects reporting up to .
What happened
OpenAI said it is improving ChatGPT for low-resource programming languages and adding evaluations for software engineering, SQLite and COBOL migration. The company reported significant coding gains across 20 programming languages and legacy software migration. It said its newest model, GPT-6.1 Sol, scored 97.9% on its COBOL test.
Why it matters
The reports describe the gains as a way to modernize old software. They give no further stakes.
Timeline
- Oct 19, 6:44 PM ET: OpenAI said it was adding evaluations for low-resource programming languages, including COBOL migration.
- Oct 19, 8:53 PM ET: OpenAI reported significant coding gains across 20 programming languages and migrating legacy software.
- Oct 20, 4:52 AM ET: OpenAI said GPT-6.1 Sol scored 97.9% on its COBOL test.
Coverage
- OpenAI Says Its Models Can Help Banks Migrate From Old COBOL Programs (engadget.com)
- GPT-6.1 Sol scores 97.9% on OpenAI’s COBOL test (Aytun Çelebi)
- OpenAI models hit 97.9% accuracy in tests to modernize old software (interestingengineering.com)
- OpenAI Adds Evaluations for Low-Resource Programming Languages (blockchain.news)
More from Technology & AI
- Eight states sue OpenAI, alleging misleading AI safety claims
- OpenAI announces solution to Navier-Stokes math problem
- Datadog bets AI agents can handle testing of AI-written code
- Anthropic and OpenAI sketch wealth-sharing plans before IPOs
- Mamdani administration proposes AI customer service rules
- Florida engineers study how AI behaves in high-risk situations