Spooky LabsPricing
Pay for what your agents use.
Inference is priced by model and token type. Compute: seconds your agent is running; a suspended agent costs nothing. No seats, no plans, no minimum. Put a card on file, buy prepaid credits, and your agents spend from that balance; when it reaches zero they stop until you top up.
Preview: billing runs in Stripe test mode. Use test card 4242 4242 4242 4242 with any future expiry and CVC. No real charge.
Rate card
Current prices
Rates from our billing catalog, in USD before tax, for the model the console offers during the preview. Your credits are drawn down at these rates.
Loading current prices…
Billing
How billing works
The gap between a meter ticking and money moving is where most billing surprises live, so here is what happens in it.
- Model-specific token prices
- Input, cached input and output tokens use the rate for the model and provider that handled the request. Thinking tokens are included in output usage. There is no single blended token price.
- Compute and storage
- Compute: seconds your agent is running; a suspended agent costs nothing. Running time is counted in seconds. There is no separate workspace-storage charge.
- Prepaid credits, and a stop at zero
- Buy a credit pack with the card on file. Tokens and running seconds draw the balance down, and when it reaches zero your agents stop until you top up; turn on auto top-up to have the platform buy the next pack at a threshold you set. Each purchase is a receipt; there is no bill at the end of the month.
- Usage takes time to appear
- The console shows the balance, the current usage estimate and the time each was retrieved. Recent usage may still be processing.
Review current usage by model and agent in the console — see what usage looks like.
Questions
Three that come up
Run one agent and read the invoice.
There is no plan to choose. Put a card on file, buy credits, start an agent, and watch the balance and its usage in the console.