Open source tools for software engineering teams using AI
← Bibliography

A benchmark for vericoding: formally verified program synthesis

Sergiu Bursuc and Theodore Ehrenborg and Shaowei Lin and Lacramioara Astefanoaei and Ionel Emilian Chiosa and Jure Kukovec and Alok Singh and Oliver Butterley and Adem Bizid and Quinn Dougherty and Miranda Zhao and Max Tan and Max Tegmark misc 2025 Source: z-spec

View original ↗

Note

LLM success rates: 82% Dafny, 44% Verus/Rust, 27% Lean. Dafny improved from 68% to 96% in one year.

Citation Key

bursuc2025vericoding