Plans & regions
Four tiers on a clean 2× ladder, three regions, one click to resize.
The plans
| Plan | vCPU | RAM | Disk | Monthly | Yearly (eff/mo) |
|---|---|---|---|---|---|
| Solo | 2 | 4 GB | 40 GB | $12 | ~$10 |
| Team | 4 | 8 GB | 80 GB | $24 | ~$20 |
| Agency | 8 | 16 GB | 160 GB | $48 | ~$40 |
| Scale | 16 | 32 GB | 320 GB | $96 | ~$80 |
Annual billing saves ~17%. Authoritative prices are always on /pricing.
What every plan includes
- Dedicated single-tenant VM — never shared kernel or container.
- Auto HTTPS via Let's Encrypt at a Nacre subdomain.
- JWT-gated web terminal.
- Nightly encrypted backups to Cloudflare R2.
- All 290+ LLM providers (BYOK).
- All messaging channels (Telegram, Discord, Slack, WhatsApp).
- User-approved OpenClaw upgrades.
- 14-day pro-rated refund on first-time paid subscriptions.
Picking a tier
- Solo
- A personal bot. One or two channels, a handful of active users per day, modest memory.
- Team
- A small team or community bot. 3–5 channels, dozens of DMs/day, more skills loaded.
- Agency
- Client work. Heavier skill libraries, faster response SLOs, several bots or personas multiplexed.
- Scale
- Production workload. Serious throughput, priority provisioning, 99.9% SLA, dedicated Slack channel.
Start one tier low — you can always upgrade in a click.
Regions
| Code | Location | Best for |
|---|---|---|
| SG | Singapore | APAC latency; our primary region. |
| US | Ashburn, VA | North America and LatAm. |
| EU | Falkenstein, Germany | EU/UK latency; GDPR residency. |
Upgrading (and downgrading)
Upgrading is an in-place Hetzner resize: the VM reboots with more vCPU, RAM, and disk. Downtime is ~30–60 seconds. Stripe pro-rates the difference to the end of your billing period.
Downgrading is trickier because disks can't shrink in place. We run a controlled migration: snapshot your/data, provision a new smaller VM, restore, swap DNS, retire the old VM. We do this for you — email support and pick a window.
Resource pressure signals
Your agent reports disk and memory usage in every heartbeat. The dashboard surfaces warnings at:
- 80% disk — yellow banner, email.
- 95% disk — red banner, email.
- 100% disk — OpenClaw is paused to prevent data corruption; email.
If you're bumping these on Solo regularly, upgrade to Team — or prune old conversation logs from the web terminal.