OpenAI reports GPT-6.1 Sol results on COBOL migration benchmark
10/19/2026 — 10/20, 15:18·2 sources (all)·2 reports (all)
Key takeaways
- OpenAI has shared results from the COBOL, Fortran and Visual Basic code migration benchmark it released on October 19. GPT-6.1 Sol passed 86…
Background & analysis
On October 19, 2026, OpenAI published a post describing its work on improving ChatGPT's code capabilities for low-resource programming languages. Alongside that work, it released new evaluations covering software engineering, SQLite, and COBOL migration, with the COBOL portion aimed at legacy system code.
The following day, an article on DEV Community reviewed the release and reported that the evaluations span code migration for COBOL, Fortran, and Visual Basic. According to that article, GPT-6.1 Sol passed 86.6% of the migration tasks at low reasoning effort, while GPT-5.6 Sol passed only 5.4% under the same setting. At high reasoning effort, GPT-6.1 Sol reached a 97.9% pass rate. The evaluation counts a migration as passing only when the output matches the original program byte for byte.
The same article noted that the test cases have not been made public, and that only one of the two evaluations can be reproduced. So far the public record stops at OpenAI's release of the evaluations and the third-party article relaying the specific scores; no further explanation of those figures from OpenAI appears in the material.
AI-generated from 2 reports · updated 15 minutes ago
Latest turnIn an October 19 post on migrating COBOL, Fortran, and Visual Basic code, OpenAI reported that GPT-6.1 Sol passed 86.6% of its COBOL migration eval tasks at low reasoning effort, versus 5.4% for GPT-5.6 Sol at the same setting, and 97.9% at high effort. A migration only counts if hidden tests produce byte-for-byte identical output, but those tests are not published and only one of the two evals can be rerun.
Related AI tools
- Heat index
- 151
- All sources
- 2
- All reports
- 2
- First seen
- 21 hours ago
Reports on this story headlines open the original
In an October 19 post on migrating COBOL, Fortran, and Visual Basic code, OpenAI reported that GPT-6.1 Sol passed 86.6% of its COBOL migration eval tasks at low reasoning effort, versus 5.4% for GPT-5.6 Sol at the same setting, and 97.9% at high effort. A migration only counts if hidden tests produce byte-for-byte identical output, but those tests are not published and only one of the two evals can be rerun.
DEV Community · AIAI score 57
OpenAI detailed its work on improving ChatGPT's coding performance in low-resource programming languages, alongside new evaluations covering software engineering, SQLite and COBOL migration. The COBOL track targets legacy codebases specifically.
OpenAI NewsFirst-partyAI score 49
Other stories people are talking about
- Heat index 94NewOpenAI launches ChatGPT Slides with real-time team collaboration1 sources
- Heat index 67OpenAI updates ChatGPT browser extension with cross-tab answers and in-page actions1 sources
- Heat index 26FTC subpoenas OpenAI and Anthropic over AI safety incidents1 sources
- Heat index 76Artificial Analysis launches HealthBench leaderboards1 sources
- Heat index 90OpenAI discloses GPT-5.6 Sol agent scanning 362 wallets for test funds2 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion