Pay per minute. No subscription, no seats.
$0.05 per billable active minute for ready-made agents, which is $3.00 per hour of an agent actually working, prorated to the second.
$0.07 per billable active minute ($4.20 per hour) for custom agents built from your own description. Building a custom agent is free.
Prepaid credit
Add $10, $25, $50, $100 through Stripe’s secure checkout. We never see or store your card number.
New workspaces get $5 of free trial credit for platform time. It isn’t money and can’t be withdrawn or refunded.
Your AI provider, billed separately
You connect your own Anthropic, OpenAI, Google Gemini or xAI API key. That provider bills its token usage to you directly. Consumer chat subscriptions don’t include API access.
Spending controls
Before each run you see its maximum charge and can lower it. The run stops at that limit. Owners can set a monthly limit for the workspace.
What’s billed
Time when a model request or tool is actually running for your task, measured to the millisecond. If an agent makes several calls at once, overlapping time counts once.
What’s not billed
- Waiting in the queue, or for your approval or input
- Retries after provider rate limits, platform errors or a worker crash
- Runs that fail because of a fault on our side
- Anything after the run finishes
Worked examples
Platform fee only. Your provider’s token charges are separate.
| Run | Active time | Platform fee |
|---|---|---|
| Short meeting-notes run | 30 seconds | $0.025 |
| Typical document summary | 1 minute | $0.05 |
| Long document, several passes | 3 minutes | $0.15 |
| 100 one-minute runs in a month | 100 minutes | $5.00 |
| Team of 3, 8 one-minute runs a day each, 21 workdays | 504 minutes | $25.20 |
A $10 top-up covers 200 active minutes. Every run has a receipt with the active time, the rate, the price version and your provider’s token usage.
Questions
- Will the price change? Each run keeps the price it started with. Changes apply only to new runs.
- Refunds? See the Refund Policy.