List your company

Dafny Code Synthesis on APPS Vericoding-derived (test)

83Pass RateModel union
35.05647.50359.9572.397May 29, 2026

Evaluation results

MethodLinks
Model unionAttempts per task=52026.0583
GPT-5 miniAttempts per task=52026.0571.6
Claude Opus 4.1Attempts per task=52026.0566.6
Gemini 2.5 FlashAttempts per task=52026.0536.9