Shared scenarios
165
Same selected cohort for both runners
Browser benchmark · aligned cohort
Biloba completes this suite 2.6× faster.
A three-run comparison of the same 165 browser scenarios, using the same application state and execution budget. Biloba averaged about 1m13s; Playwright averaged about 3m10s.
Directional · real project data
Playwright average
3m 09.88s
Range: 3m 07.57s–3m 12.07s
Biloba average
1m 12.96s
Range: 1m 11.87s–1m 13.80s
Observed speedup
2.60×
Biloba wall time ÷ Playwright wall time
Average wall time
Bars are scaled to Playwright’s average. Lower is faster.
Run-by-run timings
| Run | Playwright | Biloba |
|---|---|---|
| 1 | 3:12.07 | 1:13.21 |
| 2 | 3:10.00 | 1:11.87 |
| 3 | 3:07.57 | 1:13.80 |
| Average | 3:09.88 | 1:12.96 |
Observed spread: Playwright ±1.18% (4.49s range); Biloba ±1.32% (1.93s range). The percentage uses half the min–max range divided by the mean.
What was measured
- Workers / procs
- 5
- Retries
- 0
- App
- Fresh production build and app process per sample
- Database
- Fresh clone of the same migrated, reference-seeded PostgreSQL template
- Browser
- Chrome headless shell
- Biloba
github.com/onsi/biloba@v0.15.1- Revision
809495f
Build, app startup, and database provisioning were excluded from the measured wall time. The timings include browser, test harness, and application work.
Result quality
| Runner | Selected | Passed | Failed | Existing skips |
|---|---|---|---|---|
| Playwright | 165 | 164 | 0 | 1 |
| Biloba | 165 | 165 | 0 | 0 |
Playwright’s one skip was already present in every sample. None of the 165 selected scenarios failed in either runner.
README-ready wording
Across 165 paired browser scenarios, Playwright averaged 3m10s (3m08s–3m12s, ±1.2%) over three runs; Biloba averaged 1m13s (1m12s–1m14s, ±1.3%). Biloba completed the suite 2.6× faster in this environment.
This is a suite-level observation, not a universal framework benchmark. Hardware, browser version, parallelism, app state, and test mix can change the result.