Runware

Free hosted
OpenClaw

Spin up a sandboxed OpenClaw instance in one click, deployed from your Runware dashboard.

Your API key is wired in automatically. All the best frontier models are available on the same interface. Hosting is free. You pay only for the AI tokens your agent uses.

Deploy OpenClaw
One-click deploy · All frontier models · Fully pay-as-you-go
No subscription · $2 in free credit with a business email
Provisioning OpenClaw
  1. Provisioning your OpenClaw instance
  2. API key generated and wired in
  3. Running agents in 5 minutes
run claude-opus-4-7
Channels

OpenClaw works with the channels you already use.

Channels are configured inside OpenClaw's Control UI after deployment.

What's included

The fastest way to run OpenClaw in production.

Free
Free hosted infrastructure

Compute, storage, sandboxed container, networking. All on Runware. No monthly fee, no per-instance charge, no subscription.

1 click
One-click deploy

Sign in to Runware, click deploy, your sandboxed OpenClaw instance is live. No servers to provision, no Docker to configure, no terminal commands, no GPU provisioning.

All models
Frontier Models, One Interface

Switch between Qwen, GLM, Kimi, MiniMax, Claude, GPT-5 and Gemini from inside OpenClaw. No provider accounts to manage. No reconfiguration to swap models.

PAYG
Pay only for what you use

AI inference billed per token at Runware's published rates. Hosting is free, security and updates are included. Nothing else to pay for.

The deploy flow

From sign-up to running agent, in five minutes.

  1. 1
    Sign up to Runware
    ~ 1 minute

    No card required to start.

  2. 2
    Click "Deploy OpenClaw" from your dashboard
    1 click

    No configuration to fill in.

  3. 3
    Runware provisions your instance
    ~3 minutes

    Sandboxed container in an isolated Kubernetes namespace. Fully automatic.

  4. 4
    Your API key is wired in
    Automatic

    Pre-configured to route inference through Runware. No keys to copy, no configs to edit.

  5. 5
    Launch into OpenClaw
    Same minute

    Pick a model, connect a channel, start chatting. Both open-weight and frontier are available immediately.

Deploy OpenClaw

No manual setup · No infrastructure management · No config headaches · No local deployment · No GPU provisioning · No Docker complexity · No subscription lock-in

vs alternatives

Hosted by Runware vs the alternatives.

Self-hosting means buying a VPS, setting it up and maintaining it. Other managed providers charge $16 to $133 per month for hosting on top of whatever you pay for AI. Runware hosts everything for free and bills only for the tokens your agent uses.

  • Hosting cost
    Hosted on Runware
    Free
    Self-hosted
    $5–$20/mo VPS
    Other managed
    $16–$133/mo subscription
  • Time to deploy
    Hosted on Runware
    One click, < 5 minutes
    Self-hosted
    ~80 minutes setup
    Other managed
    A few minutes
  • Infrastructure mgmt
    Hosted on Runware
    None
    Self-hosted
    Yours
    Other managed
    None
  • Model access
    Hosted on Runware
    Open-weight + frontier, switchable
    Self-hosted
    Bring your own provider
    Other managed
    Limited or single-vendor
  • AI pricing model
    Hosted on Runware
    Pay-as-you-go
    Self-hosted
    Bring your own
    Other managed
    Often capped credits
  • Switching models
    Hosted on Runware
    One dropdown inside OpenClaw
    Self-hosted
    Reconfigure & redeploy
    Other managed
    Limited
  • Updates & security
    Hosted on Runware
    Managed
    Self-hosted
    Yours
    Other managed
    Managed
  • Sandbox isolation
    Hosted on Runware
    Included
    Self-hosted
    You build it
    Other managed
    Varies
  • Subscription lock-in
    Hosted on Runware
    None
    Self-hosted
    None
    Other managed
    Yes
Self-hosting OpenClaw, step by step
  • Provisioning a VPS15 min
  • Create SSH keys10 min
  • Connecting to the server5 min
  • Install Node.js + deps5 min
  • Installing OpenClaw7 min
  • Configuring OpenClaw18 min
  • Setting up TLS certificates10 min
  • TLS renewal cron job4 min
  • Connecting an LLM provider5 min
  • Pairing a messaging channel4 min

Total: ~80 minutes before the first message is sent. Plus ongoing maintenance, updates, monitoring and on-call.

Model access

Open-weight and frontier on the same screen.

Once your OpenClaw instance is live, switch between every model in the Runware catalogue without reconfiguring your agent. Same dashboard, same billing surface, no provider accounts to manage.

Open-weight · on Runware

Runware infrastructure

Closed-source frontier · pass-through

Market-leading rates

Full LLM model directory
Pricing

Hosting is free. AI is pay-as-you-go.

$0

Hosting

Compute, storage, sandboxed container, networking, updates and security all included at zero charge. No subscription, no monthly minimum and no per-instance fee. All you need is credit in your account for LLM usage.

Per token

AI inference

Open-weight models run on Runware infrastructure at our published rates. Closed-source frontier passed through at the provider's rate. New accounts with a business email start with $2 in free credit.

See pricing
FAQ

Questions, answered.

Under a minute from clicking deploy to a working agent. Most of the time is signing up.

Deploy your free hosted OpenClaw

Sign up. Click deploy. Start chatting.

The whole thing takes less than 5 minutes.

Deploy OpenClawRead the docs