Sandbox0 Cloud logo

Sandbox0 Cloud

Persistent sandboxes for long-running AI agents

Sandbox0 Cloud is a managed sandbox runtime for AI agents that need reusable execution environments. Agents can run code, install dependencies, manage files, and control network access through a REST API, the s0 CLI, and open-source Go, JavaScript/TypeScript, and Python SDKs. The current runtime is closed source; the historical open-source release is archived, self-hosting is no longer supported, and Bring Your Own Cloud is planned rather than available. After a successful filesystem checkpoint, the writable root filesystem retains installed tools, configuration, repositories, and user files across runtime generations. Default pause/resume restores committed files and starts fresh processes; it does not preserve live memory, sockets, or REPL sessions. Runtime-only mounts such as /tmp are excluded from filesystem persistence. Experimental memory checkpoints have separate requirements. Cloud compute costs $0.015 per GiB-hour, billed by the second using the configured memory limit. Retained incremental storage costs $0.02 per GiB-month. Pausing stops compute charges while retained data remains billable; leaving a sandbox running but idle still incurs compute charges. New users receive a one-time $5 credit without a credit card. The production API base is https://api.sandbox0.ai. SDK and HTTP automation use scoped API keys, while the CLI supports interactive device login. Verification covered the official documentation and pricing, CLI releases, SDK repositories, a public OpenAPI 3.0.3 schema, and authentication rejection for an unauthenticated sandbox-list request. Sandbox execution, checkpoint durability, and billing were not tested with an authenticated account.

Best for agent capabilities

Useful in agent tasks

Overview

Authapi key
Pricing$0.015/GiB-hour compute (billed by the second); $0.02/GiB-month retained incremental storage; $5 new-user credit.
SDKsgo, typescript, python
OpenAPIYes
MCP SupportNo

When to use Sandbox0 Cloud

Sandbox0 Cloud is listed for Sandbox, Code Execution workflows. Sandbox0 Cloud is a managed sandbox runtime for AI agents that need reusable execution environments. Agents can run code, install dependencies, manage files, and control network access through a REST API, the s0 CLI, and open-source Go, JavaScript/TypeScript, and Python SDKs. The current runtime is closed source; the historical open-source release is archived, self-hosting is no longer supported, and Bring Your Own Cloud is planned rather than available. After a successful filesystem checkpoint, the writable root filesystem retains installed tools, configuration, repositories, and user files across runtime generations. Default pause/resume restores committed files and starts fresh processes; it does not preserve live memory, sockets, or REPL sessions. Runtime-only mounts such as /tmp are excluded from filesystem persistence. Experimental memory checkpoints have separate requirements. Cloud compute costs $0.015 per GiB-hour, billed by the second using the configured memory limit. Retained incremental storage costs $0.02 per GiB-month. Pausing stops compute charges while retained data remains billable; leaving a sandbox running but idle still incurs compute charges. New users receive a one-time $5 credit without a credit card. The production API base is https://api.sandbox0.ai. SDK and HTTP automation use scoped API keys, while the CLI supports interactive device login. Verification covered the official documentation and pricing, CLI releases, SDK repositories, a public OpenAPI 3.0.3 schema, and authentication rejection for an unauthenticated sandbox-list request. Sandbox execution, checkpoint durability, and billing were not tested with an authenticated account. Consider it when an agent needs this capability through a documented service boundary instead of custom infrastructure maintained inside the application.

In 4agent's task model, the product maps to code execution. That makes it a candidate for workflows that build and deploy software.

Integration checklist

  • Access: confirm the api key authentication flow, credential scope, rotation process, and whether calls can be attributed to a user or service identity.
  • Client support: official or documented clients are listed for go, typescript, python.
  • Automation: OpenAPI is listed as available, MCP is not currently recorded, and CLI support is listed as available. Confirm each requirement in the provider's current documentation.
  • Operations: test rate limits, timeouts, retry safety, error payloads, logs, data retention, and the current paid pricing model with production-like volume.

Sources and verification

Use these primary sources to verify current behavior before making an implementation or purchasing decision. Product capabilities, limits, and prices can change after a 4agent listing is reviewed.