step-5-preview
Benchmark profile · LiveBench 2.0 ·
Results for this model
- Overall benchmark score
- 61.0 / 100
- Cost per task
- $0.7295
- Cost per success
- $1.1965
- Coding
- 85.6 / 100
- Reasoning
- 83.8 / 100
- Browser use
- 24.8 / 100
Costs are based on the benchmark’s 471 shared tasks. Cost per success adjusts the reported cost by the model’s overall score.