Browser Agent Evals
readme.md →How today’s models perform on computer use benchmarks, compared on accuracy, cost, and speed.
10 runs, ranked by accuracy
accuracy per model
accuracy
project=stagehand-devbenchmark=onlineMind2Webrows=10/10providers=3harnesses=2updated=2026-11-18T22:39:35Zrun=stagehand-evals
10 runs
accuracy vs cost per task
accuracy
| claude-opus-4-8 | hybrid | 1st84.1% | 260s | $1.156 | |
| gpt-5.5 | hybrid | 2nd72.3% | 285s | $1.730 | |
| gemini-2.5-computer-use-preview-10-2025 | cua | 3rd71.3% | 180s | 3rd$0.273 | |
| claude-opus-4-7 | cua | 66.3% | 3rd142s | $1.221 | |
| gemini-3-flash-preview | cua | 65.3% | 209s | 2nd$0.0626 | |
| claude-sonnet-4-6 | cua | 62.7% | 387s | $1.850 | |
| claude-sonnet-4-5 | cua | 57.3% | 339s | $1.498 | |
| claude-haiku-4-5 | cua | 54.0% | 213s | $0.468 | |
| gpt-5.4 | cua | 40.7% | 2nd108s | $0.474 | |
| gpt-5.4-mini | cua | 24.3% | 1st62s | 1st$0.0578 |
*these results were collected on stagehand v3 and are kept for reference. for current numbers see
Run your own evals
same harness, your models, your benchmark
Want to get your model or harness evaluated?
talk to an engineer →Your agent should be able to use a browser like you do.
The SDK for browser agents