Scrapybara
Scrapybara is a cloud service that hosts remote desktop instances (Ubuntu 22.04, Windows 11, and a lightweight Chromium browser) intended for AI agents to control. Developers provision these VMs via a Python or TypeScript SDK (or REST API) and drive them through mouse, keyboard, bash, and filesystem actions. Primary users are teams building autonomous computer-use agents, including integrations with OpenAI CUA and Claude Computer Use.
| Pricing model | Monthly subscription tiers with included compute hours; pay-per-use agent credits beyond included quota |
|---|---|
| Pricing The headline compute price as the provider states it. Sorting ranks prices stated per vCPU over time, converted to hourly. Other pricing models sort after, unranked. | $29/month (Basic) or $99/month (Pro) |
| Free while idle Whether a stopped, paused, or sleeping sandbox costs nothing. | – |
| Elastic Resources flex with what the sandbox actually uses, and the bill follows: yes when the provider bills on active or observed use, no when capacity is reserved or allocated and billed while it runs. | – |
| Memory snapshots Whether RAM state survives a pause and resume, not just disk. | – |
| Wake on request Whether a stopped sandbox wakes automatically on inbound traffic. | – |
| Start / resume Typical time from create or resume to running. Sorting ranks stated times, converted to seconds; claims with no figure sort after, unranked. | Under 1 second for Ubuntu and Browser instances; Windows instances are described as 'slow' without a specific figure |
| Max runtime The longest a sandbox may run. | Configurable via timeout_hours parameter; default 1 hour, no documented upper limit |
| Isolation The isolation technology between sandboxes. | Full Linux (Ubuntu 22.04) and Windows 11 virtual machines; open-source computer service is packaged as a Docker image with xdotool/noVNC |
| GPUs Whether GPU instances are available for sandboxes. | – |
| Docker Whether Docker containers can run inside the sandbox. | – |
| SDKs | Python, TypeScript |
✓ yes · ✕ no · – no cited public fact.
The homepage now says “Scrapybara built computers for agents. Now we're building Capy, the AI software engineer,” while the documentation still describes the remote desktop instances. Ubuntu has fast startup and 1x compute cost; Windows has slow startup and 2x compute cost. The API exposes a `warm_pool` status whose behavior and billing the documentation does not explain, and Scrapybara publishes no standalone per-instance compute rate.
Sources
- https://github.com/Scrapybara/scrapybara-oss/tree/main/computer
- https://docs.scrapybara.com/ubuntu
- https://docs.scrapybara.com/windows
- https://scrapybara.com/blog/sdk2
- https://docs.scrapybara.com/sdk-reference/python
- https://github.com/scrapybara
- https://docs.scrapybara.com/browser
- https://docs.scrapybara.com/best-practices
- https://scrapybara.com/
- https://docs.scrapybara.com/
- https://www.ycombinator.com/launches/MHN-scrapybara-a-computer-for-your-ai
- https://docs.scrapybara.com/cursor-rules
- https://docs.scrapybara.com/api-reference/start.md
Last verified 2026-08-22. Corrections land through the update log.