INSTALL OPENCLAW
Subscription Tiers

Install OpenClaw Subscription Plans

Connect your own server, bring your preferred AI credentials, and let us manage the rest. Every plan includes all messaging platforms, automatic updates, and full data sovereignty.

Prerequisites

Two things to prepare before signing up

LLM Provider Access

  • Anthropic Claude API key
  • OpenAI API key
  • OpenRouter account
  • Or self-hosted Ollama / llama.cpp

A Linux Machine with SSH

  • Any VPS or dedicated server from a supported provider
  • Or bare metal hardware you control

Supported Infrastructure

DigitalOceanVultrHetznerAWSAkamaiContaboOVHHostingerAWS LightsailAzureGoogle CloudOracle CloudScalewayCustom Server

Base

$29/mo

server hosting billed separately

Everything one person needs to run an AI agent on a single box.

  • Up to 1 managed server
  • Unlimited agentsActual capacity depends on your server hardware
  • All platform connections
  • Core usage metrics
  • Email support
  • Teams
  • Custom domain
  • Activity logging
  • Custom API
Choose Base

Pro

$49/mo

server hosting billed separately

Multi-server management with analytics and team collaboration.

  • Up to 10 managed servers
  • Unlimited agentsActual capacity depends on your server hardware
  • All platform connections
  • Analytics dashboard
  • Priority email & chat support
  • 1 team (up to 5 members)
  • Custom domain
  • Activity logging
  • Custom API
Choose Pro
Recommended

Business

$149/mo

server hosting billed separately

Full-featured platform access with audit trails and API extensibility.

  • Up to 50 managed servers
  • Unlimited agentsActual capacity depends on your server hardware
  • All platform connections
  • Advanced analytics & reporting
  • Priority support
  • 5 teams (up to 25 members)
  • Custom domain
  • Activity logging
  • Custom API
Choose Business

Agency

$399/mo

server hosting billed separately

White-label ready with unlimited capacity and hands-on support.

  • All Business features, plus:
  • Unlimited managed servers
  • Unlimited teams & members
  • Dedicated account manager
  • Complete activity logging
  • White-label features
  • Pre-built agent configurations
Choose Agency

Platform Integrations

Messaging channels included with every plan

WhatsAppTelegramDiscordSlack

Requirements

Minimum server specifications

Every managed server must meet these baseline specs. Heavier workloads and concurrent agent sessions benefit from additional resources. GPU hardware is recommended for local LLM inference.

CPU

2 vCPU cores minimum

Memory

2 GB RAM minimum

Storage

20 GB disk minimum

AI Providers

Bring your own AI credentials

Supply API keys from any of the following services. You keep full control over which models run on your infrastructure and how much you spend on inference.

Anthropic Claude

Deep analysis, long-context reasoning, and reliable code output

OpenAI

Versatile GPT models covering text, vision, and function calling

OpenRouter

Proxy layer that routes requests to any compatible model API

Ollama

Run Llama, Mistral, and other open models on your own hardware

llama.cpp

Lightweight C++ runtime for CPU and GPU inference without containers

FAQ

Pricing questions answered

Where do AI credentials come from?
You bring them. Create an account with Anthropic, OpenAI, or OpenRouter and paste your API key into the Install OpenClaw dashboard. Alternatively, point the agent at a locally running Ollama or llama.cpp instance — no external key required. Messaging-platform tokens (WhatsApp Business API, Telegram Bot API, etc.) are also configured by you.
What happens on my server during setup?
After you share SSH credentials through the dashboard, our provisioning system connects once to install the OpenClaw runtime, apply security hardening, and configure automatic updates. From that point the management layer keeps the instance healthy without manual intervention. Nothing leaves your server — logs, files, and model outputs stay local.
How much hardware does OpenClaw need?
The baseline is 2 CPU cores, 2 GB of memory, and 20 GB of storage on any mainstream Linux distribution. That is enough for cloud-based LLM calls. If you intend to run open-weight models locally through Ollama, allocate a GPU with at least 8 GB of VRAM.
What counts as a single gateway?
One gateway equals one OpenClaw identity — its own workspace directory, configuration file, and credential set. A single gateway can bridge several messaging channels at once (e.g., WhatsApp and Discord simultaneously), so you rarely need more than one per use case.
Are plan changes instant?
Upgrades activate within seconds; your invoice is prorated for the remaining days in the current cycle. Downgrades queue for the next renewal date so you keep full access until then. No fees or waiting periods apply to either direction.
Do I have to sign a contract?
No binding term exists on any tier. Monthly subscribers cancel with one click and keep access until the period expires. Annual subscribers lock in a 20 percent discount and follow the same cancellation flow — access continues through the pre-paid year.
How do I pay?
Checkout runs through Stripe. We accept all major card networks and PayPal out of the box. Subscribers on the Business or Agency tier can request PDF invoices with NET-30 terms by contacting our finance team.

Questions before you commit?

Reach out and we will match you with the tier that fits your infrastructure and team size.

Get in TouchBrowse Documentation