Alternatives to Alibaba Cloud AgentBay

An open-source Kubernetes controller for managing stateful agent sandboxes, with lifecycle controls, persistent storage, and warm pools.

Persistent cloud sandboxes for running commands, hosting services, and managing agent workspaces through an instance API.

Cloud computers for agents with programmatic command execution, file operations, and machine lifecycle controls.

Managed all-in-one sandboxes combining code execution, filesystem access, and browser automation for agent applications.

Managed sandbox environments for agents to execute code, analyze data, and work with files through Amazon Bedrock AgentCore.

Remote machines for agent execution, with command and file access across managed compute and machines connected from your own infrastructure.

Isolated Linux sandboxes for running agent commands and working with files through Python and TypeScript SDKs.

Cloud virtual machines for agent execution, with public compute pools and private pools for customer infrastructure.

Managed code interpreter sessions for running untrusted code and exchanging files through a REST API, with Hyper-V isolation.

Managed sandbox groups for agent workloads, with sandbox lifecycle, command execution, snapshots, and suspend/resume controls.

Remote sandboxes for agent command execution and file operations, with an SDK integration for agent frameworks.

Cloud sandbox containers with process, filesystem, networking, and snapshot controls for agent workloads.

Managed sandbox VMs for running agent-generated code, working with files, and retaining environment state between sessions.

Managed Linux machines for running commands, moving files, and managing persistent agent workspaces through SDKs and a CLI.

Cloud Linux virtual machines with command execution, filesystem access, and lifecycle controls through SDKs, an API, and a CLI.

Remote execution environments for shell commands and code, with files, snapshots, and pause/resume support.

Cloud Linux sandboxes for agent code execution, with files, persistent state, and lifecycle controls through SDKs, an API, and a CLI.

Managed Linux computers with API access to command execution, files, and machine lifecycle controls.

Isolated environments accessed from Cloudflare Workers for executing generated code and workloads that should run separately from the application.

Managed computers with APIs for terminal commands, files, browser operations, and direct desktop actions.

Cloud microVMs for command execution, with persistent files and processes, forks, and suspend/resume controls.

Managed microVM sandboxes for command execution, files, snapshots, and persistent agent workspaces.

Managed sandbox computers for running agent-generated code, with programmatic process execution, file access, and persistent snapshots.

Isolated Linux sandboxes for agent code execution and file operations, managed through an application SDK.

Cloud virtual machines for agent workloads, with command execution, SSH access, and lifecycle management through an API and CLI.

Linux microVMs on Deno Deploy for running untrusted code, executing commands, and managing files from an SDK or API.

Cloud execution workspaces with APIs for running commands, managing files, and controlling sandbox lifecycle.

Isolated environments for AI coding agents, available on your own machine or Docker-managed cloud compute.

Sandbox infrastructure that gives agents isolated Linux environments for running code, working with files, and pausing or resuming sessions.

Cloud Linux virtual machines for agent workloads, with command execution, file access, persistent workspaces, and pause/resume controls.

Managed sandboxes for executing agent-generated code, with stateful sessions and lifecycle controls through Google Cloud SDKs.

Cloud machines built on Hugging Face Jobs for running commands and transferring files from Python or the CLI.

Cloud machines for agent workspaces, with CLI, SDK, and API access to commands, files, SSH, and suspend/resume controls.

Cloud sandboxes for stateful Python and shell execution, with file operations through an application SDK.

Cloud CAD desktops for agents, with Python SDK access to desktop actions, application sessions, and code execution.

Cloud sandboxes with command execution, files, port forwarding, and configurable lifecycle policies through an API, SDK, or CLI.

Cloud microVMs for agent command execution, with persistent filesystem state and hibernate/resume lifecycle controls.

Persistent cloud development machines for agent work, controlled over SSH with workspace creation, command execution, and lifecycle operations.

Ephemeral execution environments on Koyeb, with SDKs for creating sandboxes, running processes, and accessing files.

Isolated execution environments for agents to run commands, access files, and expose services through LangSmith.

Cloud sandboxes with APIs for commands, files, interpreter sessions, SSH, snapshots, and desktop control.

Remote execution environments for agent code, with SDK and CLI access to sandbox lifecycle and commands.

Persistent cloud machines for agent workspaces, with command-line access to machine provisioning and lifecycle controls.

A local microVM runtime for running untrusted agent workloads and code through a CLI or an embedded SDK.

Cloud microVM sandboxes with API access to command execution, files, previews, forks, and pause/resume controls.

Programmatically created sandbox environments on Modal for executing commands and running untrusted or generated code.

Cloud virtual machines for agent workspaces, with programmatic command execution, snapshots, and instance lifecycle controls.

Managed Linux sandboxes for running agent commands and managing files through an API, SDK, or CLI.

Managed execution environments for agent code and file operations, with programmatic sandbox lifecycle controls.

Managed Linux execution environments for agent commands and files, controlled through a CLI or MCP tools.

Sandbox compute for agents on Northflank’s cloud or customer infrastructure, with APIs for deployment and lifecycle management.

Managed execution environments for agents to run code, install dependencies, use browsers, and preserve state between sessions.

Managed Firecracker sandboxes for command execution, interpreter sessions, and file access through a sandbox API.

Self-hosted sandbox infrastructure with a control plane and Linux VM workers for commands, files, snapshots, and agent workspaces.

Self-hosted containerized execution servers with APIs for remote workspaces, shell commands, and file transfer.

Cloud sandboxes for agent code execution and file operations, with SDK-based lifecycle management.

Self-hosted sandbox infrastructure for agent code execution, command and file operations, with local Docker and Kubernetes deployment options.

Cloud computers for agents with APIs for desktop control, shell commands, files, and computer lifecycle management.

Managed Linux sandboxes with API and SDK access to command execution, files, and lifecycle controls.

MicroVM sandboxes for agent execution, with SDKs and a CLI for commands, files, snapshots, and forks.

Hosted Linux containers for agent workspaces, with APIs and CLI commands for remote execution and persistent files.

Managed agent sandboxes with isolated runtimes for code, files, and browser tasks, plus templates and session persistence.

Isolated microVM workspaces for agent coding, evaluations, and experiments, with custom images and filesystem checkpoints.

Managed Linux sandboxes for agent command execution, with SDK controls for files, checkpoints, and sandbox lifecycle.

An embeddable execution runtime with isolated filesystems, processes, terminals, and language tools, also available as managed durable VMs.

Managed sandbox development environments for software agents, with reusable environment templates and snapshot-based session persistence.

Cloud execution environments for agents, with APIs, SDKs, and a CLI for commands, files, and runtime lifecycle management.

Cloud Linux sandboxes with APIs, SDKs, CLI commands, and MCP tools for execution, files, and lifecycle management.

Cloud virtual machines for agent workloads, with SDK access to commands, files, checkpoints, and persistent execution state.

Managed sandbox environments with a REST API for creating machines, executing commands, and managing their lifecycle.

Hosted agent sandboxes with persistent writable filesystems, command execution, and checkpoint-based pause and resume.

Self-hosted Kubernetes sandbox infrastructure with warm pools, an E2B-compatible API, and command-line platform management.

Managed sandbox environments for agent code execution and file operations, controlled through a Python SDK.

Cloud Linux workspaces for agent execution, with SDK access to commands, files, and workspace lifecycle controls.

Persistent cloud microVMs for agent workspaces, controlled through SSH commands for machine creation, execution, and files.

Linux microVMs for local or managed cloud execution, with SDK, API, and CLI controls for commands, storage, and machine lifecycle.

Cloud execution environments for commands and stateful code, with SDK access to files, snapshots, and forks.

Persistent Linux environments for agents, with command execution, filesystem checkpoints, and automatic idle handling.

Managed container environments for agent code execution, with API and SDK controls for sandbox lifecycle and files.

Firecracker sandbox infrastructure with a managed API and a deployable control plane for running agent code.

Cloud machines for agent execution, including GPU workloads, with APIs, SDKs, and a CLI for commands, SSH, snapshots, and forks.

Cloud Linux sandboxes with SDK access to code execution, processes, files, terminals, and lifecycle management.

Managed code, browser, and custom-image sandbox environments for agent applications.

Self-hosted microVM sandbox infrastructure with an E2B-compatible API, snapshots, and single-node or clustered deployment.

Managed Linux sessions for agent execution, with SDK access to commands, snapshots, and persistent volumes.

Isolated sandbox environments for agent harnesses, tool execution, coding workflows, and builds, with suspend and resume support.

Hosted execution environments created from Docker-based snapshots for agent tool calls, generated code, and evaluation workloads.

Cloud execution environments for agent code and tools, with programmatic sandbox creation and command execution.

MicroVM sandboxes for running commands and transferring files through a JavaScript SDK, available in managed cloud or private deployments.

Cloud boxes for agent workspaces, with SDK access to commands, files, and optional coding-agent execution.

Managed Linux microVMs for running agent-generated code, executing commands, managing files, and serving application previews.

Managed container sandboxes for agent code, browsers, and computer-use workloads, with API, SDK, and MCP access.

Managed execution of agent-generated JavaScript and Python, with SDK integrations for running code and working with files in sandboxed environments.

Cloud browser infrastructure for agents, with programmatic session control and browser authentication workflows.

Managed browser sessions that an application or agent can launch and control through its own code over CDP.

Managed browser infrastructure for creating, controlling, and observing browser sessions used by AI agents.

Browser infrastructure for running automation with tools such as Playwright and Puppeteer, available as a managed service or a Docker deployment.

Managed computers with APIs for terminal commands, files, browser operations, and direct desktop actions.

Managed cloud desktops for computer-use agents, provisioned through an SDK or CLI with configurable fleet environments.

Cloud browser sessions for agents, with navigation, screenshots, browser profiles, and live viewing through SDK and API controls.

Cloud CAD desktops for agents, with Python SDK access to desktop actions, application sessions, and code execution.

Cloud browser infrastructure that agents can control through browser automation protocols or computer controls, with session viewing and replay tools.

Persistent Linux desktops for agents, exposed through MCP with screenshots, desktop actions, and user-scoped computer sessions.

Cloud computers for agents with APIs for desktop control, shell commands, files, and computer lifecycle management.

Browser infrastructure with a session API for agent-driven web automation, offered through Steel Cloud or a self-hosted browser server.