Managed OpenClaw Hosting
Cloudways OpenClaw Hosting: Plans, Pricing, Setup & Sizing
Cloudways OpenClaw Hosting is designed for people who want an always-on OpenClaw agent without manually building the server that keeps it alive. Cloudways provisions the managed infrastructure, deploys the agent, applies validated updates, provides backups and gives you dashboard and terminal access. You then connect your own model provider, configure the agent, add the channels you want and decide which tools or external services it may use.
The key question is not whether OpenClaw can run on a cheaper VPS. It can. The useful question is whether you want to own the operating work around it and, if you choose Cloudways, which instance is large enough for your actual workflows. This page focuses on that buying intent: plan selection, realistic cost, channel load, sub-agent headroom and the point where Scout stops being the sensible choice.
Interactive tool
Cloudways OpenClaw Server Size Calculator
Estimate the smallest sensible Cloudways instance from workload, concurrency, channels, sub-agents and runtime pattern. The result is a planning recommendation, not a benchmark guarantee.
Quick answer
Which Cloudways plan should you use for OpenClaw?
Use Scout for testing, light personal use and one or two simple OpenClaw workflows. Operator is the safer default for an agent that is active every day, especially when it is connected to messaging channels or performs regular browser, file and scheduled tasks. Squad is better when several workflows or sub-agents operate in parallel, while Swarm is intended for heavy, continuous use with many agents and tools.
Cloudways currently lists the same four standard instance tiers for OpenClaw and Hermes, so the plan decision comes from workload rather than a different OpenClaw price table. Your LLM usage is not included in the server price. Use the calculator below to size the infrastructure, then budget model-provider charges separately.
Cloudways OpenClaw hosting plans
OpenClaw uses the standard Managed AI Agent tiers. The important differences are CPU, memory, storage and bandwidth, not a different feature bundle for each agent.
| Plan | Compute | Storage / bandwidth | Standard price | Best starting fit |
|---|---|---|---|---|
| Scout | 1 vCPU / 2 GB RAM | 50 GB SSD / 2 TB bandwidth | $9.99/mo standard | Light checks, testing and one or two simple workflows |
| Operator | 2 vCPU / 4 GB RAM | 80 GB SSD / 3 TB bandwidth | $19.99/mo standard | An active agent doing focused, regular work |
| Squad | 4 vCPU / 8 GB RAM | 160 GB SSD / 5 TB bandwidth | $39.99/mo standard | Multiple workflows or coordinated sub-agents |
| Swarm | 8 vCPU / 16 GB RAM | 320 GB SSD / 6 TB bandwidth | $79.99/mo standard | Many tools, agents and continuous production activity |
Current standard pricing
Cloudways Managed AI Agent instance plans
LLM provider usage is billed separately.
- 1 vCPU
- 2 GB RAM
- 50 GB SSD
- 2 TB bandwidth
- 2 vCPU
- 4 GB RAM
- 80 GB SSD
- 3 TB bandwidth
- 4 vCPU
- 8 GB RAM
- 160 GB SSD
- 5 TB bandwidth
- 8 vCPU
- 16 GB RAM
- 320 GB SSD
- 6 TB bandwidth
Why managed OpenClaw hosting is different from installing it on a VPS
Running OpenClaw yourself gives you maximum control, but it also makes the server part of your job. You choose the Linux image, secure access, configure services, manage process restarts, monitor disk space, install updates, protect credentials and maintain backups. None of those tasks are conceptually difficult for an experienced administrator, yet together they create a maintenance queue that continues long after the first successful installation.
Cloudways removes much of that queue. Its managed OpenClaw product emphasizes isolated infrastructure, automatic security updates, validated OpenClaw updates, automated backups with one-click restore, dashboard access and terminal access. That combination is aimed at users who want the benefits of a persistent agent but do not want the agent to become another production server they have to babysit every week.
The trade-off is abstraction. You are choosing the environment Cloudways supports rather than designing every layer yourself. For many users that is a benefit because it narrows the number of things that can go wrong. For teams with unusual kernel requirements, custom networking or a deeply standardized container platform, self-hosting may remain more appropriate. The value question is therefore operational responsibility, not whether one approach is technically possible.
What OpenClaw is good at on an always-on server
OpenClaw makes sense as an always-on agent because its usefulness increases when it can remain reachable, retain context and respond through the channels where you already work. Instead of launching a local process for each task, the hosted gateway can sit behind WhatsApp, Slack, Discord or Telegram and continue running scheduled jobs, monitoring work or handling requests even when your laptop is closed.
That persistent model works well for broad assistant tasks: checking information, watching builds, working with files, running commands, triggering recurring jobs, sending notifications and coordinating other agents. The specific tools you enable determine the real capability. A hosted OpenClaw instance is not valuable merely because it is online; it becomes valuable when the connected skills and services map to repetitive work you would otherwise do manually.
Before adding many tools, define a small set of high-value workflows and measure them. An OpenClaw agent that reliably handles three repetitive tasks is more useful than one connected to twenty systems with unclear permissions. This approach also improves sizing because you can observe how real workflows affect memory, CPU and storage before adding more channels or sub-agents.
How much RAM OpenClaw needs on Cloudways
There is no honest single RAM number for every OpenClaw deployment because the workload is shaped by concurrent tasks, helper processes, connected channels, browser activity, local files and sub-agents. Cloudways therefore offers tiered instances rather than a one-size server. Scout provides 2 GB RAM, Operator 4 GB, Squad 8 GB and Swarm 16 GB, with CPU and storage increasing alongside memory.
For a personal agent that receives occasional messages and performs simple checks, 2 GB can be a rational starting point. Once the agent is active through the day, 4 GB gives more breathing room for overlapping jobs and background processes. If you expect multiple sub-agents, several simultaneous workflows or heavier browser and terminal activity, 8 GB becomes easier to justify. A 16 GB Swarm instance should be chosen because the workload is persistently heavy, not because the name sounds safer.
The calculator deliberately treats concurrency and sub-agents as stronger signals than user count. One person can drive a busy agent, while a small team may use an agent lightly. After deployment, watch the instance over a representative period. If memory pressure or latency is consistently high, scale. If one brief burst is the only issue, do not automatically double the plan before understanding what caused it.
WhatsApp, Slack, Discord and Telegram can increase concurrency
OpenClaw becomes particularly attractive when connected to messaging channels because the agent can be used from places where people already communicate. Cloudways documents WhatsApp, Slack, Discord and Telegram in its Managed AI Agents workflow. The convenience is obvious: a user can message the agent without opening another dashboard. For a team, the agent can also become a shared interface for repetitive information or operations.
Every extra channel creates another path for work to arrive. That does not mean each channel consumes a fixed amount of server capacity, but it can increase overlapping sessions, event processing and background activity. If you intend to connect several channels and allow multiple people to use them, do not size the server as if it were a private dashboard-only agent. Operator or Squad can be a better starting point even when the underlying prompts are not computationally complex.
Channel security matters as much as capacity. Keep Slack tokens, Telegram bot tokens and other credentials private, review the permissions each integration requests and decide whether the agent should respond to every message or only explicit commands. If a channel can invoke tools that change external systems, define confirmation rules for destructive or paid actions. The easiest interface should not automatically have the broadest authority.
OpenClaw model choice and the true cost of ownership
Cloudways uses a bring-your-own-model-key model. That means you can connect a supported provider such as OpenAI, Anthropic or Google, but the model provider bills usage separately. The $9.99 Scout price is therefore not the complete monthly cost of the agent. A busy workflow using a premium model and large contexts can generate model charges that exceed the server price by a wide margin.
The upside is flexibility. You can change models or route different tasks to different models without moving the hosting layer. Simple classification or extraction can use a lower-cost model while complex reasoning is reserved for a stronger one. Good prompt design, smaller context windows and deliberate scheduling can reduce API spend without changing the Cloudways plan. Infrastructure sizing and model economics should be optimized independently.
When evaluating the current Cloudways Managed AI Agents promotion, apply any discount only to the infrastructure component unless the offer explicitly says otherwise. Model providers are not part of the Cloudways invoice. For a realistic business case, calculate standard Cloudways cost, promotional Cloudways cost if eligible, expected LLM spend and any external paid services the agent will call.
Backups and validated updates reduce an awkward maintenance risk
Agent software changes quickly, and that speed can be uncomfortable on an always-on system connected to real tools. A self-hosted update may introduce a dependency problem, change behavior or require manual rollback. Cloudways highlights validated OpenClaw updates rather than blindly applying every upstream release, which is an important distinction for users who prefer a managed change process over being first to every new version.
Automated backups and one-click restore address the other side of the problem. If a configuration, skill or update leaves the agent in a bad state, recovery should not begin with reconstructing a server from memory. Backups do not replace source control for your own code or safe handling of external credentials, but they reduce the infrastructure recovery work when the hosted environment itself needs to be restored.
For production use, decide what should exist outside the server as well. Keep critical prompts, skill definitions, integration setup notes and credential inventories in an appropriate system of record. A platform backup helps recover the instance; operational documentation helps recover the workflow. The strongest setup uses both rather than assuming a snapshot can explain how the agent was intended to behave.
When Scout is enough and when to move to Operator, Squad or Swarm
Scout is enough when OpenClaw is genuinely light: occasional messages, a small number of simple workflows, limited local state and little simultaneous activity. It is an excellent way to validate whether the agent is useful before paying for unused headroom. Starting small also forces you to see which workloads actually matter instead of designing capacity around hypothetical future automation.
Move to Operator when the agent becomes part of daily work. Regular channel activity, scheduled jobs, browser or terminal tasks and overlapping requests make 4 GB RAM and 2 vCPUs a more comfortable baseline. Move to Squad when multiple sub-agents or workflows run together often enough that concurrency is normal. Swarm becomes rational when many agents or tools run continuously and the cost of queueing or resource contention matters more than the higher monthly price.
Cloudways supports scaling, so the first choice is not permanent. Measure before and after a change and keep a note of why you scaled. If a single badly behaved workflow causes the pressure, fixing that workflow may be better than buying more resources. If the workload itself has grown, scaling is the correct response. The calculator is designed to give you a starting hypothesis, not replace real monitoring.
A practical OpenClaw buying checklist
Before paying for an OpenClaw instance, define the first three workflows you expect it to perform and the channels that will trigger them. Then decide whether those workflows are occasional, daily or continuous and whether any can overlap. This simple inventory makes the plan decision much easier because you are sizing for real activity rather than the vague idea of having an always-on AI assistant.
Next, list every credential the agent will need: the LLM key, messaging tokens, MCP credentials and any application-specific API keys. Decide which of those credentials can be read-only, which can make changes and who is responsible for rotation. An OpenClaw deployment can be technically reliable and still be badly designed if every connected service receives broad permissions for convenience.
Finally, decide what success looks like after the first month. Measure completed workflows, time saved, failure rate, model spend and server utilization. If OpenClaw is useful but resource-constrained, scale the instance. If the server is mostly idle but the workflows are valuable, keep the smaller tier. If the workflows do not create value, a larger plan will not fix the underlying problem.
Methodology and primary sources
Cloudzat bases plan resources, pricing, deployment requirements and supported workflow claims on current Cloudways product and help-center documentation. Last verification date: August 26, 2026.
- Cloudways Managed AI Agents getting-started guide
- Cloudways Managed AI Agent instance sizing guide
- Cloudways OpenClaw managed hosting
- Cloudways Hermes managed hosting
- Cloudways Managed AI Agents general availability announcement
- Cloudways communication channels guide
- Cloudways MCP connection guide for Managed AI Agents
Frequently asked questions
How much does Cloudways OpenClaw hosting cost?
Cloudways currently publishes standard Managed AI Agent prices of $9.99 for Scout, $19.99 for Operator, $39.99 for Squad and $79.99 for Swarm. Model-provider charges are separate.
Does Cloudways manage OpenClaw updates?
Cloudways markets validated OpenClaw updates and automatic security updates as part of its managed product. That reduces routine maintenance, although you should still review important workflow changes and keep your own operational documentation.
Can OpenClaw on Cloudways connect to WhatsApp?
Yes. Cloudways documents WhatsApp as a supported communication channel for Managed AI Agents and provides a QR-code pairing workflow for OpenClaw. Slack, Discord and Telegram are also documented options.
Is 2 GB RAM enough for OpenClaw?
It can be enough for light personal use, testing and one or two simple workflows. If the agent is busy every day, has several channels, uses sub-agents or runs overlapping jobs, Operator or a larger tier is a safer starting point.
Does the Cloudways price include OpenAI usage?
No. You provide your own supported LLM API key, and the model provider bills usage separately from Cloudways infrastructure.
Can I scale OpenClaw later?
Yes. Cloudways documents upscaling and downscaling Managed AI Agent instances. Evaluate usage over time rather than reacting to one temporary spike before changing plans.
Cloudzat may earn a commission if you sign up for Cloudways through links on this page. This does not change your price. Managed AI Agent features, integrations, plan resources, model-provider costs and promotions can change, so confirm critical details in Cloudways before purchase or production deployment.