How to Deploy OpenClaw on Render (The Honest Guide)

OpenClaw is a self-hosted AI agent that runs 24/7 on a machine you control. It connects to messaging apps — Telegram, Discord, Slack, WhatsApp — and can take real actions: send emails, create GitHub issues, browse the web, run shell commands. You message it, it does the thing.

The catch: it needs a machine to run on. This guide covers how to host it on Render, including every mistake I ran into.

Why Render

There are three real options for hosting something like this:

Raw cloud (AWS, GCP, Azure) — you rent a virtual machine and manage everything yourself. OS updates, security, networking, scaling. Maximum control, maximum friction. Fine if you have a DevOps background. Overkill if you just want an AI agent running.

Platform-as-a-Service (Render, Railway, Fly.io) — you give them your code or a Docker image and they handle the machine underneath. You think about apps, not servers. This is the right tier for most people.

No-code / managed (Vercel, Netlify) — highly opinionated, designed for specific use cases like static sites. Not flexible enough for something like OpenClaw.

Render sits in the PaaS tier. It's the "modern Heroku" — same idea, better pricing, native Docker support. Why Render over the other PaaS options:

Railway is very similar but more experimental. Render is more stable and has better documentation.
Fly.io gives you more networking control, but it requires more configuration. Good if you need multi-region. Overkill otherwise.
DigitalOcean has a 1-click OpenClaw Marketplace deploy, which is simpler — but you're managing a VPS, not a PaaS. More control, more responsibility.
Oracle Cloud Free Tier gives you 4 cores and 24GB RAM permanently for free. The tradeoff: availability in good regions is spotty, and setup is more DIY.

Render wins for this use case because: one-click Blueprint deploy, persistent disk built-in, auto-HTTPS, and the OpenClaw team maintains an official render.yaml that handles all the configuration automatically.

Understanding the Plans

This is the most important thing to understand before you start, because the wrong plan will waste your time.

Plan	Cost	RAM	Spin-down	Persistent Disk
Free	$0	512 MB	After 15 min idle	❌ Not available
Starter	$7/mo	512 MB	Never	✅ 1 GB+
Standard	$25/mo	2 GB	Never	✅ 1 GB+

The free tier has two fatal problems for OpenClaw:

No persistent disk — every redeploy wipes your config, plugins, and API keys.
512 MB RAM — OpenClaw's Node.js process needs at least 1 GB to start reliably. On 512 MB it crashes with a heap out-of-memory error every time.

Starter ($7/mo) has one problem: 512 MB RAM. Same OOM issue. You can work around it with NODE_OPTIONS=--max-old-space-size=400, but it's right at the edge and will fail under any real load.

Standard ($25/mo) is what actually works. 2 GB RAM, never spins down, persistent disk. OpenClaw's own docs recommend 2 GB. This is the one to use.

If you want to experiment first before paying $25/mo: use Starter, add the memory workaround, and accept it might be flaky. Upgrade to Standard when you want it to actually work.

Step 1: Deploy with the Blueprint

OpenClaw has an official Render Blueprint — a single config file that defines your entire server setup. You don't configure anything manually.

Go to render.com and create an account if you don't have one.
Click this deploy link: render.com/deploy?repo=https://github.com/openclaw/openclaw
You'll be prompted to set one value: SETUP_PASSWORD. This is the password you'll use to access your OpenClaw dashboard. Pick something you'll remember.
Click Apply and wait. The first deploy takes 3–5 minutes — it's building a Docker image.

Your service URL will follow the pattern https://openclaw-xxxx.onrender.com.

Step 2: Fix the Instance Type

Before anything else, go to your service settings and upgrade the instance.

In the Render Dashboard, click your openclaw service.
Go to Settings → find Instance Type → click Update.
Select Standard (2 CPU, 2 GB RAM).
Save.

If you skip this, your service will crash on every deploy with:

FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory

Step 3: Set the Node.js Memory Limit

Even on Standard with 2 GB, you need to tell Node.js it's allowed to use that memory. By default Node caps itself at ~256 MB.

Go to your service → Environment tab.
Find NODE_OPTIONS (or add it if it doesn't exist).
Set the value to: --max-old-space-size=1536
Click Save Changes. Render will trigger a redeploy automatically.

This tells Node it can use up to 1.5 GB of heap, leaving ~500 MB for the OS and container overhead on a 2 GB machine. This is exactly what OpenClaw's official Fly.io docs recommend for equivalent-size machines.

Step 4: Complete the Setup Wizard

Once your deploy shows as Live in the Render Dashboard:

Navigate to https://your-service.onrender.com/setup
Enter the SETUP_PASSWORD you set in Step 1.
Select your model provider (Claude/Anthropic or OpenAI).
Paste your API key.
Optionally connect messaging channels — Telegram, Discord, Slack.
Click Run Setup.

Your control panel will be at https://your-service.onrender.com/openclaw.

Step 5: Install the Composio Plugin (Optional)

Composio gives OpenClaw access to 1,000+ third-party tools — Gmail, GitHub, Notion, Slack, Linear, Salesforce, and more — through a single plugin. Instead of writing agent code yourself, OpenClaw becomes the agent.

First, get your consumer key from dashboard.composio.dev.

Then in the Render Dashboard → your service → Shell tab, run:

openclaw plugins install @composio/openclaw-plugin
openclaw config set plugins.entries.composio.config.consumerKey "ck_your_key_here"
openclaw gateway restart

Once it loads, all Composio tools are registered directly into the agent. You can message OpenClaw things like "create a GitHub issue and send me a Slack summary" and it just works.

Troubleshooting

Deploy fails with TypeScript errors during build:plugin-sdk:dts

This is a known issue where broken commits land on main. It's not your Render config — it's a bug in OpenClaw's source code. Check openclaw/openclaw issues to see if there's an open bug for it. Use the Rollback button in Render's Events tab to go back to the last working deploy while you wait for a fix.

Service crashes with heap out of memory

You're on a plan with insufficient RAM. 512 MB is too small. Upgrade to Standard and set NODE_OPTIONS=--max-old-space-size=1536.

Config disappears after redeploy

You're on the free tier, which has no persistent disk. Upgrade to Starter or Standard.

Service is slow to respond the first time

If you're on the free tier, the service spins down after 15 minutes of inactivity. First request after spin-down takes a few seconds to cold start. Upgrade to Starter or Standard for always-on.

Keeping It Updated

Render won't auto-deploy when OpenClaw releases updates. To update manually:

Go to your service → click Manual Deploy → select Deploy latest commit.

You can also export your full config anytime at /setup/export — downloads a portable backup you can restore on any OpenClaw host.

Follow me on Instagram for more: @shawnbuilds