The three-way buying decision
This shortlist starts with AWS AgentCore Browser and Lightpanda, then includes Notte as a third implementation path for budgeted browser fleets. A team has a recurring browser budget and a target number of completed jobs. Plan capacity, billing rounding and failed attempts affect what the budget buys.
Choose a plan that can run the workload before calculating usage. Retry traffic, idle sessions and rounded minimum time can change the cost per completed job.
A proposed acceptance case
A team has a recurring browser budget and a target number of completed jobs. Plan capacity, billing rounding and failed attempts affect what the budget buys.
This is an evaluation design, not a claim that this workflow has been executed successfully on every provider.
| Part of the test | Required observation |
|---|---|
| Controlled input | A representative job volume, attempted session durations, concurrency requirement and network usage estimate. |
| Pass condition | The eligible configuration stays within the modeled budget while completing the required work, with each usage meter and plan credit applied once. |
| Negative case | Reprice the workload with additional failed attempts and idle time. The model must expose the higher cost instead of preserving an optimistic cost-per-success figure. |
| Evidence to retain | Attempted and completed jobs, rounded browser usage, separate meters, plan allowance and estimated invoice. |
Documented mechanisms, side by side
These labels describe product mechanisms. The linked source and the acceptance case determine whether a mechanism meets this workflow's requirement.
| Requirement | AWS AgentCore Browser | Lightpanda | Notte |
|---|---|---|---|
| Browser control | Browser automation endpoint | Compatibility-dependent | Documented |
| Saved browser state | Explicit state design | Not established | Documented |
| Live view | Documented | Check engine support | Documented |
| Deployment choice | AWS service | Engine available | Enterprise options |
Sources: docs.aws.amazon.com [1] ↗lightpanda.io [2] ↗docs.notte.cc [3] ↗docs.notte.cc [4] ↗docs.notte.cc [5] ↗docs.notte.cc [6] ↗
Pricing model
AWS AgentCore Browser: A complete AWS workflow may include browser use, the controlling compute service, storage and other metered services. Price the selected architecture and region instead of treating a browser benchmark rate as the final invoice.
Lightpanda: Use the cloud billing unit and account limits for hosted access. A lower rate only helps if the target pages and required actions complete correctly on the selected engine.
Notte: The public usage rates are $0.01 per browser-hour, $0.05 per Function-hour and $10 per GB of managed proxy traffic. Paid plans return their price as usage credits. Price the relevant meters and plan eligibility for the workload.
Sources: aws.amazon.com [1] ↗lightpanda.io [2] ↗notte.cc [3] ↗
Browser control
AWS AgentCore Browser: Create or select a Browser Tool, then start an isolated browser session. The documented default session timeout is 15 minutes and the maximum is eight hours. Browser automation uses the session's WebSocket endpoints.
Lightpanda: Lightpanda Cloud supplies browser access for automation. Connect using the documented cloud path and evaluate compatibility with the control library and site behavior needed by the workflow.
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.
Sources: docs.aws.amazon.com [1] ↗lightpanda.io [2] ↗docs.notte.cc [3] ↗
Deployment ownership
AWS AgentCore Browser: Use an AWS-managed Browser Tool or configure a custom browser with recording, networking and an execution role. Keep the controller's deployment separate in the architecture and operational estimate.
Lightpanda: Compare hosted Lightpanda access with operating the engine yourself. The engine's supported APIs and website compatibility remain relevant in either deployment.
Notte: Use a remote browser from your existing process or deploy the browser automation as a Notte Function. These choices put lifecycle, scheduling and monitoring responsibilities in different places.
Sources: docs.aws.amazon.com [1] ↗lightpanda.io [2] ↗docs.notte.cc [3] ↗
A common browser-usage scenario
For an illustrative fleet of 10 browsers active eight hours a day over 30 days, browser usage is 2,400 hours. The table multiplies that same usage by the saved Arena rate. It excludes other providers' plan minimums, allowances and rounding, so it is not a total-invoice comparison.
Task completion below comes from the separate throughput workload. Preserve its date and failure context; a low usage rate or fast lifecycle does not establish that an authenticated business job completes. BrowserArena is maintained by Notte.
| Provider | Arena rate / browser-hour | Browser usage at 2,400 hours | Throughput task completion |
|---|---|---|---|
| AWS AgentCore Browser | $0.1273 | $305.52 | 51% on 2026-10-17 |
| Lightpanda | $0.06 | $144.00 | 100% on 2026-10-18 |
| Notte | $0.01 | $24.00 | 91% on 2026-10-18 |
Sources: browserarena.ai [1] ↗browserarena.ai [2] ↗aws.amazon.com [3] ↗lightpanda.io [4] ↗notte.cc [5] ↗
What should change the choice?
Notte's public $0.01 browser-hour rate makes it a useful cost baseline. Apply its paid-plan minimum and any proxy, execution or identity usage before estimating the actual recurring bill.
AWS AgentCore Browser: AgentCore Browser's AWS integration can matter more than moving a browser endpoint. Retain AWS identity and observability requirements in the evaluation, and separately account for the code runner and website login before comparing Notte's integrated workflow.
Lightpanda: Lightpanda introduces a browser-engine decision. Keep it on the shortlist when its supported actions fit the real pages you use; a successful CDP connection or text task does not establish visual or full-browser equivalence with the proposed Notte workflow.
Included allowances, a different billing minimum or an existing commercial commitment can make retaining another provider reasonable. Compare invoices for eligible configurations, not hourly labels alone.
Sources: aws.amazon.com [1] ↗docs.aws.amazon.com [2] ↗lightpanda.io [3] ↗lightpanda.io [4] ↗notte.cc [5] ↗docs.notte.cc [6] ↗docs.notte.cc [7] ↗
How to remove an option from the shortlist
For this workflow, require the following evidence from AWS AgentCore Browser, Lightpanda and Notte: Attempted and completed jobs, rounded browser usage, separate meters, plan allowance and estimated invoice.
Remove a proposed configuration when its documented mechanism or a controlled pilot cannot meet the pass condition. Keep an undocumented feature as an unresolved question instead of inventing an equivalence. Do not select a provider solely because it appears in this shortlist.
- The eligible configuration stays within the modeled budget while completing the required work, with each usage meter and plan credit applied once.
- Reprice the workload with additional failed attempts and idle time. The model must expose the higher cost instead of preserving an optimistic cost-per-success figure.
Sources: aws.amazon.com [1] ↗lightpanda.io [2] ↗notte.cc [3] ↗
Make a bounded pilot
Price a representative week from attempted sessions and actual duration, then compare it with the plan's included credits and limits.
Use the same authorized account, input and final-state check in each configuration. Treat a missing product contract as an open question; do not turn an undocumented capability into an unsupported claim or a successful test result.
- Capture source attempted usage and plan allowance for a representative period.
- Model the same work using the target plan, rounding rules and separate service meters.
- Canary a bounded budget before changing the full fleet's billing path.