OpenClaw comes preinstalled on Ubuntu 24.04. Connect your model account, configure your channels, and give your assistant a private server for ongoing tasks.
An entry profile for evaluating an API-backed gateway with a small agent workspace.
Deploy NowOur recommended starting profile for a personal assistant using an external model service.
Deploy NowRoom to evaluate browser-assisted workflows alongside the gateway and supporting processes.
Deploy NowMore working memory for concurrent jobs and larger workspaces, sized from measured demand.
Deploy NowA larger memory and storage budget for several configured agents and their retained files.
Deploy NowA high-memory profile with 1,280 GB storage for workflows retaining larger datasets and artifacts.
Deploy NowOur highest listed OpenClaw profile for workloads needing more RAM, compute, and workspace storage.
Deploy NowCompare memory, compute, storage, and monthly prices for each plan.
| Plan details | OpenClaw VPS 2 | OpenClaw VPS 4 | OpenClaw VPS 8 | OpenClaw VPS 16 | OpenClaw VPS 32 | OpenClaw VPS 64 | OpenClaw VPS 96 |
|---|---|---|---|---|---|---|---|
| Memory | 2 GB RAM | 4 GB RAM | 8 GB RAM | 16 GB RAM | 32 GB RAM | 64 GB RAM | 96 GB RAM |
| Shared compute | 1 vCPU | 2 vCPU | 4 vCPU | 6 vCPU | 8 vCPU | 16 vCPU | 24 vCPU |
| Storage | 55 GB NVMe | 80 GB NVMe | 160 GB NVMe | 320 GB NVMe | 640 GB NVMe | 1280 GB NVMe | 1600 GB NVMe |
| Bandwidth | 2 TB | 3 TB | 4 TB | 5 TB | 6 TB | 10 TB | 15 TB |
| Dedicated IP | 1 included | 1 included | 1 included | 1 included | 1 included | 1 included | 1 included |
| Operating system | Ubuntu 24.04 | Ubuntu 24.04 | Ubuntu 24.04 | Ubuntu 24.04 | Ubuntu 24.04 | Ubuntu 24.04 | Ubuntu 24.04 |
| Software | OpenClaw preinstalled | OpenClaw preinstalled | OpenClaw preinstalled | OpenClaw preinstalled | OpenClaw preinstalled | OpenClaw preinstalled | OpenClaw preinstalled |
| Server access | Administrator | Administrator | Administrator | Administrator | Administrator | Administrator | Administrator |
| Your next step | Choose plan ↗ | Choose plan ↗ | Choose plan ↗ | Choose plan ↗ | Choose plan ↗ | Choose plan ↗ | Choose plan ↗ |
Swipe to compare every plan →
The application is installed for you. You connect your own model account and decide which channels and tools your assistant can use.
The gateway, browser processes, and local files use VPS resources. External model usage has its own service limits and charges.
| Your workload | Plan to evaluate | What to check |
|---|---|---|
| Light trials | 2 GB / 1 shared vCPU | Try simple tasks with a small workspace. Check memory pressure and response time before relying on ongoing automation. |
| Personal assistant | 4 GB / 2 shared vCPU | Our recommended starting point for a personal setup. Leave headroom for the Gateway, integrations, and operating system. |
| Multiple channels | 8 GB / 4 shared vCPU | Evaluate with your actual channels and browser tools. Watch peak memory and CPU when tasks overlap. |
| More concurrent work | 16–32 GB | Choose from measured demand for jobs, supporting processes, and larger workspaces. A memory tier does not guarantee an agent count. |
| Large retained workspaces | 64–96 GB | Review working memory, retained files, and backup size. Set retention rules so logs and artifacts do not fill the disk. |
Model inference runs at your chosen external service. These plans list no GPU allocation; local-model inference needs a separate capacity review. Compare all plan resources ↗
The Gateway holds workspace state and integration credentials. Choose how you will reach it and which services it can contact.
Administrator access gives you control of the server. You configure the assistant's accounts, integrations, and permissions.
Limit integration credentials to the required work, and review skills and plugins before enabling their tools. File and command tools can affect the environment they run in. Keep application and editor credentials private.
OpenClaw documents sandboxing as off by default. Check the installed policy and verify isolation for enabled tools, including reachable files and credentials.
Understand OpenClaw's trust boundary ↗Keep the OS and OpenClaw updated, retain protected backups of state and workspace data, and test recovery. External model APIs, subscriptions, software licences, and other third-party services are separate from the VPS.
Plan an off-server backup ↗Comparing agent workflows? Explore our planned Hermes Agent VPS hosting for its memory, reusable skills, and scheduled-task guidance. Hermes ordering is coming soon.
Explore regional VPS hosting, or talk to our team about your application requirements.
Browse VPS locations ↗Start Live ChatA private VPS can run your OpenClaw assistant, scheduled tasks, and connected tools without keeping your own computer running.
OpenClaw is preinstalled on Ubuntu 24.04. You configure your model account, channel connections, and tool permissions after deployment; external accounts and usage are separate.
Open the application URL supplied with your service and sign in with its application credentials. Allow the initial setup to finish if the page is not yet available. Configure a model connection, send a test message, then connect one channel and restrict access to intended users. The setup guide on this page explains the sequence.
The listed plans cover the VPS resources. External model APIs, subscriptions, software licences, and other third-party services are separate.
You have administrator access to configure your software, integrations, and permissions. Keep backups and protect the credentials used by connected tools.
Start with 4 GB RAM and 2 shared vCPU for a personal assistant using an external model service; evaluate 8 GB RAM and 4 shared vCPU for multiple channels. The 2 GB plan has 1 shared vCPU and suits light trials. Browser processes, concurrent jobs, and retained files can require more capacity. Monitor representative workloads before choosing a larger tier. No GPU allocation is listed; local-model inference needs a separate capacity review.
Use the supplied HTTPS application URL and application credentials for browser access. Keep the underlying Gateway and administrative services protected by authentication and firewall rules. Allow the connections required by your chosen model and channels; follow each channel guide for any webhook or pairing requirements.
Preinstallation provides the application. You still choose which accounts, skills, and tools it may use. Review the installed execution policy and verify sandbox settings before enabling file or command tools. OpenClaw documents sandboxing as off by default; check the actual deployment rather than assuming isolation is enabled.
Each plan includes the monthly bandwidth allowance shown in its pricing card and the resource comparison. The allowance measures data transferred per month. Monitor outbound data for your workload, and confirm any additional-usage charges with AwakeHost before ordering.