# Autonomous Browser Tasks API

One implementation from Browser Use, paid per call at the provider's price.

## Implementations

- [Browser Use autonomous browser task](/providers/browser-use/run-task.md): `browser-use/run-task`, usage based, the maximum is shown before a run. Give a natural-language task to a cloud browser agent (Browser Use v4). Needs task.

## Supporting tools

- [Browser Use task result](/providers/browser-use/get-run.md): `browser-use/get-run`, no charge per call. Status, result text, error and cost of a Browser Use run (queued, dispatching, running, completed, failed, cancelled). Needs run id.
- [Cancel a Browser Use task](/providers/browser-use/cancel-run.md): `browser-use/cancel-run`, no charge per call. Cancel a Browser Use run in progress. Needs run id.

## Other capabilities

- [Cloud browser sessions](/tools/browser-automation/cloud-browser.md): 4 tools, from $0.002 per minute.

## Providers

- [Browser Use](/providers/browser-use.md): 3 tools.

## Questions

### Which tool should my agent use?

Your agent compares the implementation by price and inputs with inspect_tool and proposes one before it runs.

### How is it billed?

At the cost the provider reports for each call, with no markup. Your agent sees the maximum before it runs a tool.

[Browser automation](/tools/browser-automation.md) · [Connection guide](/connect.md) · [Billing](/docs/billing.md)

---

[HTML page](/tools/browser-automation/browser-tasks) · [Agent index](/llms.txt)