Browser Agent Evals
readme.md →How today’s models perform on computer use benchmarks, compared on accuracy, cost, and speed.
15 runs, ranked by accuracy
accuracy per model
accuracy
project=stagehand-devbenchmark=hardbenchrows=15/57providers=4harnesses=7updated=2026-09-03T01:42:51Zrun=stagehand-evals
| gpt-6-astra | codex | 1st87.0% | 3rd320s | $3.073 | |
| claude-fable-5-1 | claude code | 2nd84.0% | 552s | $0.925 | |
| claude-opus-5 | claude code | 3rd78.0% | 678s | $1.967 | |
| gpt-5.6-sol | codex | 74.0% | 346s | 1st$0.502 | |
| gpt-5.6-sol | eve | 74.0% | 1st276s | $0.856 | |
| claude-opus-5 | deep agents | 74.0% | 567s | $1.502 | |
| claude-opus-5 | eve | 74.0% | 552s | $1.809 | |
| gpt-5.6-sol | fx | 71.0% | 392s | 3rd$0.622 | |
| muse-spark-1.3 | mastra | 71.0% | 557s | $2.085 | |
| gpt-5.6-sol | deep agents | 71.0% | 349s | 1st$0.502 | |
| grok-4.6 | deep agents | 71.0% | 559s | $2.955 | |
| claude-opus-5 | fx | 71.0% | 735s | $10.023 | |
| gpt-5.6-sol | mastra | 71.0% | 2nd303s | $1.412 | |
| grok-4.6 | mastra | 71.0% | 1154s | $3.902 | |
| grok-4.6 | cursor | 71.0% | 668s | $3.296 |
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