SettledModels

OpenAI says GPT-6 Astra scores 87.4% on a 20-language coding benchmark

OpenAI says GPT-6 Astra averages 87.4% on its MultiLangBench, up from 26.8% for GPT-5.6 Sol, with large gains in COBOL, Fortran, Visual Basic and OCaml, per Developer Tech News and Thurrott. It is available in the API. The benchmarks are OpenAI's own, and OpenAI's post could not be opened for this entry.

HEAT
OUTLETS: 9
FIRST SEEN: Oct 19, 2026, 08:00 ET
LAST SIGNAL: 5d ago

Primary source

Thurrottthurrott.com/a-i/openai-a-i/343049/openai-launches-a-coding-model-for-older-programming-languages

Coverage · 10 articles, oldest first

Monday, October 19, 2026

08:00
OpenAIPrimary
Advancing Coding in Low-Resource Programming Languages

Tuesday, October 20, 2026

06:10
Digit
digit.in/news/general/openai-adds-real-time-collaboration-to-chatgpt-slides-introduces-other-new-features.html
06:10
Tech in Asia
techinasia.com/news/openai-models-improve-cobol-fortran-work
06:10
RuntimeWire
runtimewire.com/article/openai-low-resource-language-coding-benchmarks
06:10
Crypto Briefing
cryptobriefing.com/openai-upgrades-model-cobol-fortran-legacy-code/
07:40
Developer Tech News
developer-tech.com/news/openai-reports-87-4-multilangbench-pass-rate-gpt-6-astra/
10:10
Thurrott
thurrott.com/a-i/openai-a-i/343049/openai-launches-a-coding-model-for-older-programming-languages
12:40
Portland Oregon startups ...
siliconflorist.com/2026/10/20/silicon-florist-links-arrangement-for-october-20-2026/
14:40
BleepingComputer
bleepingcomputer.com/news/artificial-intelligence/openai-says-gpt-61-sol-is-dramatically-better-at-migrating-legacy-code/
17:10
Gizmodo
gizmodo.com/openai-says-its-models-can-give-the-financial-sector-a-coding-makeover-2000828398

ShareX LinkedIn Hacker News Email

← The whole wire