Step 5 Preview

Benchmark Model Rank Results
agents-on-agents-last-exam-ale-cliStep 5 Preview (High)–Pass Rate: 29.5
code-generation-on-deepsweStep 5 Preview (High) (SWE-agent)–Pass@1: 0.677
code-generation-on-programbenchStep 5 Preview (High)–Average Test Pass Rate: 80.5
code-generation-on-terminal-bench-4-0Step 5 Preview (High)–Accuracy: 33.3