Browser Benchmarks
Headless browser session benchmarks — create, connect, navigate, and release latency across providers. 100 iterations per run.
Performance Over Time
Composite Score
Detailed Metrics
Provider | Score | Total (med) | Create (med) | Connect (med) | Navigate (med) | Success |
|---|---|---|---|---|---|---|
| Tilion | 98.8 | 0.13s | 0.01s | 0.04s | 0.06s | 100% |
| Browser Use | 96.2 | 0.44s | 0.15s | 0.16s | 0.09s | 100% |
| Browserbase | 94.8 | 0.66s | 0.14s | 0.25s | 0.15s | 100% |
| Hyperbrowser | 94.6 | 0.67s | 0.20s | 0.15s | 0.21s | 100% |
| Notte | 85.4 | 1.65s | 0.34s | 0.42s | 0.73s | 100% |
| Steel | 79.9 | 2.07s | 0.70s | 0.83s | 0.13s | 99% |
| Kernel | 41.6 | 0.22s | 0.02s | 0.04s | 0.12s | 43% |
Want to see a provider added?
Methodology
What We Measure
Each benchmark provisions a headless browser session, connects over CDP, navigates to a target page, and releases the session, measuring the time for each phase independently.
Five metrics are captured per iteration: create, connect, navigate, release, and total end-to-end latency.
Each provider is tested with 100 iterations per run. Benchmarks run automatically via GitHub Actions on a recurring schedule. All results are committed to the public benchmarks repo.
How We Score
The Composite Score is a weighted blend of total-latency percentiles and create latency, multiplied by the success rate. Latency is scored against a fixed 10-second ceiling: 100 × (1 − value / 10,000ms).
The weighted score is multiplied by the success rate (0–1), so providers that fail frequently are penalized proportionally.
- • Total median: 40%
- • Create median: 30%
- • Total P95: 20%
- • Total P99: 10%