Pricing
Pay for what your agent uses. Cap what it can.
Compute and AI metered as dollar credits. Pick a plan for the included allowance, set a spend cap per key, and let your agents run. No surprise bills.
Free / Trial
Kick the tires.
$0
Start free
- $2 (trial) included credits
- Server: Shared, sleeps fast
- 1 GB workspace
- 1 live app
- Community support
Starter
One agent, real work.
$29/mo
Choose Starter
- $10 included credits
- Server: 1 vCPU / 1 GB
- 5 GB workspace
- 1 live app
- Email support
Pro
Ship full-stack apps.
$99/mo
Choose Pro
- $40 included credits
- Server: 2 vCPU / 4 GB
- 20 GB workspace
- 5 live apps
- Priority support
Scale
Many agents, committed use.
Custom
Contact us
- Committed included credits
- Server: Custom
- Custom workspace
- Unlimited* live apps
- SLA support
| Per agent server | Free / Trial | Starter | Pro | Scale |
|---|---|---|---|---|
| Server | Shared, sleeps fast | 1 vCPU / 1 GB | 2 vCPU / 4 GB | Custom |
| Persistent workspace | 1 GB | 5 GB | 20 GB | Custom |
| AI credits / mo | $2 (trial) | $10 | $40 | Committed |
| Live public apps | 1 | 1 | 5 | Unlimited* |
| Egress / mo | 10 GB | 100 GB | 250 GB | Custom |
| AI overage | Hard stop | ~1.4× base rate | ~1.4× base rate | Negotiated |
| Egress overage | n/a | $0.12 / GB | $0.10 / GB | Negotiated |
| Support | Community | Priority | SLA | |
| Per-key spend cap |
How usage works
Simple, metered, and capped
Credits are just dollars.
1 credit = $1 of metered usage. Each plan includes a monthly allowance. No tiers to decode, no token math, just dollars.
Charged to the API key.
No plan field at provision time. Agents authenticate with a key and usage is billed to it. Limits come from the account behind the key.
You set the ceiling.
A spend cap per key. Hit it and the agent stops spending, no surprise bills, no autonomous overspend.
FAQ
Questions, answered
What's a credit?
A credit is $1 of metered usage (compute + AI). Each plan includes a monthly credit allowance; you can see usage in real time on your dashboard.
What happens when I run out?
Paid plans roll into overage at ~1.4× the base rate. The Free plan hard-stops. Either way, your per-key spend cap is the hard ceiling.
Can an agent spend money on its own?
Agents provision servers and incur usage on their own, that's the point. But they can only spend up to the spend cap you set on their API key.
How do agents provision servers?
Over MCP (the create_server tool) or a REST call authenticated by the agent's API key. There's no plan field and no human approval in the request.
Do servers persist?
Yes. Each agent's server keeps its workspace and state between tasks, and idle servers scale to zero so you're not paying for nothing.
Can I run many agents?
Yes. Issue an API key per agent (or share one), each with its own spend cap. Usage rolls up to your account.