Low Latency · Multi-Instance · AI Automation

Virtual phone platform built for AI

Powered by our proprietary Android virtualization engine, it provides a virtual phone environment for AI agents, AI workflows, and mainstream AI tools, with batch execution and programmable automation.

Free Download

Windows

Download

macOS

Apple Silicon Intel

Linux

Download

Core Advantages

Why Choose VMOS Edge

Software-hardware integrated design deploys virtual Android environments locally for a low-latency, high-stability cloud phone experience.

Zero-Latency Local Compute

Cloud phones run on local hardware — no remote server needed. Responses are faster and the experience is close to a native device.

Proprietary Virtualization Engine

Supports Android 10 ~ 16 running in parallel with deep kernel optimizations, delivering performance on par with physical devices.

Data Security & Privacy

All data is stored on a local SSD. You retain full data sovereignty — nothing is uploaded to any cloud server.

Centralized Management

The PC client enables batch management, screen mirroring, and API integration across multiple devices and virtual machines.

AI Automation

AI-Driven Automation Capabilities

AI Agents Directly Operate Cloud Phones

Hand complex cloud-phone tasks straight to the AI Agent. You describe the goal, and it understands the interface, plans the steps, completes the actions, then sends the key result back.

Assign Tasks in Natural Language

Describe the goal and expected result, and the AI breaks the task into executable steps.

Observe While It Executes

It keeps reading the live screen and key elements, adjusting actions as the interface changes.

Adapts to UI Changes

It can keep going through dialogs, jumps, and branching flows based on the current screen.

Results Return Automatically

When the task is done, it returns screenshots, summaries, and key results for quick follow-up.

AI Workflows Generate Reusable Automation Scripts

The built-in Coding Agent turns natural-language chat into automation scripts. After generation, you can keep editing, debugging, and saving them so one request becomes a reusable workflow.

Describe the Task

State the goal, target screens, and expected result, and the AI understands what needs to be done.

AI Generates the Script

It generates an executable script covering common actions like launching apps, navigating, clicking, and typing.

Debug While Testing

Run it immediately and adjust as needed to verify the logic, parameters, and execution order.

Save as a Standard Workflow

Once it works, save it as a reusable template for repeat tasks and easier team sharing.

Product Capabilities

Powerful Product Capabilities

Universal AI Tools

CLI

Claude CodeGemini CLIOpenAI Codex CLIQwen CodeKimi CLIiFlow CLIQoder CLI

Editors / IDEs

CursorVS Code (Copilot)WindsurfClineRoo CodeAugment CodeContinueZedTrae (ByteDance)CodeBuddy (Tencent)

Desktop / Others

ChatGPT DesktopGoose AIOpenCodeFactory DroidOpenClawOther MCP Clients

MCP

Connection layer

skills

Reusable actions

Universal AI Tool Integration

Works with general AI tools such as Claude Code, Codex, Cursor, and Gemini, using MCP and skills to connect VMOS Edge controls without locking into a single client.

Connect directly with mainstream AI toolsMCP Guide Skills Install Guide

Agent Demo

AI is operating

For example: open Weibo and summarize today's trending searches

AI is operating the cloud phone

Trending list detected

Weibo Hot SearchTop 10 detected

Holiday travel topic keeps rising

Summary ready

3 topics extracted and key points grouped

TaskUnderstand what data should be collected

ViewLocate the trending list and key titles

RunSummarize the hot topics and return the result

Open Weibo

Read trends

Return summary

AI Agents

No fixed script is needed in advance. Just describe the task goal, and the AI can recognize the interface while completing search, click, input, and verification steps on the cloud phone, with results returned in real time.

Say the goal directlyAI acts liveResults return live

Automation Script Studio

Built-in Coding Agent

Task request

Open Settings, enter Developer Options, and enable USB debugging.

Chat input

AI script generation

AI auto-generates an editable script

Understand goal

Generate steps

Fill params

Generated result

3 runnable steps

enable-usb-debugging.workflow

app.launch("settings")

settings.open("developer_options")

toggle.enable("usb_debugging")

AI Workflows

Built-in Coding Agent lets you generate automation scripts through natural-language chat, then keep editing, debugging, saving, and reusing them at scale.

Natural-language chatAI-generated scriptsDebug and reuse

One Box, Multiple Cloud Phones

A single VMOS L1 can run multiple independent Android VMs simultaneously. Each instance has its own system environment, storage, and network configuration with full resource isolation.

Unified PC Management

Batch-manage all devices and VMs from the desktop client — real-time screen mirroring, file transfer, ADB debugging. One person can oversee hundreds of cloud phones.

API Control Console

HTTP API · 192.168.1.10:8088

POST

/activity/start

{ "package_name": "..." }

Start target app

POST

/input/click

{ "x": 540, "y": 1280 }

Inject touch event

GET

/screenshot/format

format=webp

Return screenshot

Start / Stop

Tap / Input

Shell

Screenshot

Programmable Automation

REST API and JavaScript SDK let you control every VM instance precisely: start/stop, push files, take screenshots, inject touch events, execute shell commands — from quick scripts to production automation.

Full developer docs and API referenceDeveloper Guide

Hardware Specs