What you pay (not us)
| Component | Cost | Pays to |
|---|---|---|
| OpenClawMU software | $0 | Free. Apache-2.0. |
| LLM API calls | Per-token, varies | Anthropic / OpenAI / your provider |
| Compute | $$$ per month for a VM | Your cloud provider (AWS, Hetzner, your closet) |
| Storage | ~pennies / GB / month | Your storage provider |
| Channel APIs | Mostly free, some metered | WhatsApp, Slack, etc. |
What we don't take
- No per-seat fees. Run as many tenants as your hardware can handle.
- No per-message fees. The gateway doesn't meter you.
- No "Enterprise" upsell. SSO integration, audit logs, sandboxes — all in the OSS release.
- No telemetry. The gateway phones nowhere except the LLM provider you've configured.
What you might pay us, optionally
Two things, both opt-in, both via direct contract:
- Hosted ops (planned, not yet GA) — we run an OpenClawMU gateway for you, on hardware you control via Tailscale. Useful if you want the product without operating a Node process.
- Support contracts — response-time SLAs, dedicated engineer hours for integration work, security review. Email contact@neullabs.com if interested.
Why this pricing
Multi-tenant AI infrastructure is going to be a critical primitive in the next 18 months. Critical primitives belong in the commons. The closed-source bot platforms charge what they charge because their leverage is the chokepoint they own. The open-source bet is that more usage, more contributions, and a healthier ecosystem outweighs anything we'd extract via licensing.
We'd rather have ten thousand people running OpenClawMU than a thousand paying us for it.