StepCodeBench - internal release score
Code
Step's internal coding benchmark, with unreported cohort and grading. Diagnostic.
Top models (higher is better)
| Model | Score |
|---|---|
| Step 5 Preview | 49.0 |
Benchmarks
218 sources · 1,060 models × 2,130 benchmarks · 109,661 scoresCode
Step's internal coding benchmark, with unreported cohort and grading. Diagnostic.
| Model | Score |
|---|---|
| Step 5 Preview | 49.0 |