How much time does the browser add to a task?
A session creation number measures one API operation. A usable browser also needs a connection and a navigation, and a long-running workflow pays for many actions after startup. Compare lifecycle latency and running-task throughput separately.
A workflow with one navigation and a hundred browser actions may favor a different execution path than a workload that opens and closes a browser for every action. Test both shapes before making a platform-wide speed claim.
Compare the browser lifecycle
Kernel: Kernel exposes browser creation and control APIs. Its documentation distinguishes on-demand browsers from configured browser pools, so compare the acquisition path your application will actually use.
Hyperbrowser: A session gives your code a WebSocket endpoint for CDP-compatible automation. The response also provides a live-view path. Session configuration controls which browser features the run uses.
Sources: kernel.sh [1] ↗hyperbrowser.ai [2] ↗
Compare the evidence from a run
Kernel: Live browser URLs support an embedded view and a read-only option. Replays and browser telemetry are separate observability surfaces; app invocation logs add execution context.
Hyperbrowser: Hyperbrowser documents rrweb-based web recordings and MP4 video recordings. These capture different artifacts. Enable the appropriate mode before the run and keep the recording associated with the session.
Sources: kernel.sh [1] ↗hyperbrowser.ai [2] ↗
Compare deployment ownership
Kernel: Kernel Apps move the automation process next to the browser. Alternatively, an external process can control a browser through the browser API. Choose based on execution locality and deployment ownership.
Hyperbrowser: Control a Hyperbrowser session from your application using the supported SDK or CDP library. Treat sandbox hosting as an additional execution choice and check its own runtime and storage contract.
Sources: kernel.sh [1] ↗hyperbrowser.ai [2] ↗
How to evaluate this choice
For short tools, lifecycle overhead can dominate. For a browser that stays open, action latency, task completion and drift matter more. Select the measurement that resembles your application.
- Use the same target page, browser configuration and controller workload.
- Record the runner and browser regions, account plan and concurrency.
- Keep the measured sample count and any warmup or retry behavior visible.
- Calculate tail percentiles from per-session totals, not sums of phase percentiles.
- Classify configuration failures before interpreting them as product failures.
Published browser measurements
These are the latest provider runs in the saved BrowserArena snapshot checked on October 18, 2026. Dates and runner environments can differ. A lifecycle test opens, connects, navigates and releases a browser; the throughput test runs a separate browser action sequence. Neither measures managed authentication or deployed Function reliability.
Use these observations to choose what to test in your application. An account-configuration failure, an incompatible page action and a timed-out navigation require different interpretations. BrowserArena is maintained by Notte and publishes its source and methodology.
| Provider | Lifecycle median | Lifecycle P95 | Actions / second | Run dates |
|---|---|---|---|---|
| Kernel | 334 ms | 700 ms | 3.32 | 2026-10-18 / 2026-10-17 |
| Hyperbrowser | 878 ms | 944 ms | 2.03 | 2026-10-18 / 2026-10-17 |
Sources: browserarena.ai [1] ↗browserarena.ai [2] ↗github.com [3] ↗
Which approach fits?
Choose Kernel when your code makes many browser calls and you want to deploy it beside the browser, or you need Kernel's managed sign-in flow.
Kernel tradeoff: Browser-local code changes the deployment model. Check app action behavior, account limits and persistent-state ownership before moving an external worker into it.
Choose Hyperbrowser when you need existing Playwright, Puppeteer or Selenium code to use a managed browser with saved state and named credentials.
Hyperbrowser tradeoff: Browser sessions, sandbox execution and action caching are separate surfaces. A cached action sequence is not the same deployment contract as an independently versioned browser Function.
Sources: kernel.sh [1] ↗hyperbrowser.ai [2] ↗
A useful acceptance check
Write a pass/fail result for the real task before running the evaluation. Record the account, configuration and code version next to that result. Repeat failures need an explanation, not an automatically different browser configuration.
When Notte belongs on this shortlist
Kernel's browser-local Apps make execution placement part of the decision. Compare the repeated action path with the proposed Notte controller or Function deployment before concluding that a lifecycle or hourly-rate difference decides the workflow.
Hyperbrowser's profiles, credential tools and Action Caching deserve separate evaluation. A recorded action sequence and a deployed Python handler have different maintenance contracts, so decide which one your team will edit after the website changes.
Notte: Create an owned browser session, connect Playwright or Puppeteer over CDP, and stop the session when the work finishes. Browser lifecycle management remains necessary even when the control socket disconnects.
Consider this third option when its documented mechanism addresses a requirement left open by the two providers above. A working configuration on either original provider can still be the better choice; compare the task and ownership cost before switching.
Sources: kernel.sh [1] ↗hyperbrowser.ai [2] ↗docs.notte.cc [3] ↗