One balance.
Three products.
Add credit once. Requests, reviews and hosting all draw from it, all count toward one budget, and all land in Logs with what they cost. No seats, no contracts.
The provider’s price, plus a 7.5% fee.
Your charge settles from the cost each request reports, so cache discounts and provider prices pass straight through. Same fee on every model, no minimum.
Pay for the work, not for seats.
Reviews and automation runs bill as requests on your Auditor key at Gateway prices. A small pull request usually costs a fraction of a cent. Cap it per month and per pull request.
One plan for the team.
Allowances are pooled across every app, and development is free. Upgrades take effect now, prorated; downgrades on the first of next month.
Inference BureauApps plans | |||
|---|---|---|---|
| Included each month, pooled across apps | |||
Runs Actions, webhooks, background jobs and sync passes | 50k Apps pause past this | 500k then $20 per 1M | 5.00M then $20 per 1M |
Requests Pages and API calls through your apps’ hostnames | 100k Apps pause past this | 1.00M then $15 per 1M | 10.00M then $15 per 1M |
Storage Records, vectors, files and the event log | 0.50 GB Apps pause past this | 5.0 GB then $1 per GB-month | 50 GB then $1 per GB-month |
Bandwidth What your apps send | 5.0 GB Apps pause past this | 50 GB then $0.40 per GB | 500 GB then $0.40 per GB |
Models and embeddings The ai client and vector fields | Gateway price Like any model request | Gateway price Like any model request | Gateway price Like any model request |
| Apps | |||
Production apps | 1 | Unlimited | Unlimited |
Development environments One per app, never billed | Unlimited | Unlimited | Unlimited |
Custom domains Across the team’s apps | — | 5 | 25 |
Development servers Live dev server with hot reload; awake at once | 2 Up to 100 awake hours per app a month | 10 Fair use, about 100 h per app a month | 25 Fair use, about 100 h per app a month |
| Team | |||
Members | Unlimited | Unlimited | Unlimited |
Logs kept | 1 day | 14 days | 30 days |
Restore records back | 1 day | 7 days | 30 days |
At your limit What happens when you reach it | Apps pause Until next month or an upgrade | 72-hour grace period Production keeps running | 72-hour grace period Production keeps running |
| On every plan | |||
Hosted Next.js apps on your own hostname | |||
Data models, actions and vector fields | |||
Webhooks, jobs and integrations | |||
API, MCP and the ib-apps CLI | |||
Questions people ask.
How the one balance behaves when it runs low, what the budget counts, and what you get on paper.
What happens when the balance runs out?
Requests and new reviews stop with an error that says why. Apps on Pro and Scale get 72 hours of grace, with production still serving. Top up, or turn on auto top-up, and everything resumes within a minute.
Does one budget cover all three?
Yes. The team’s monthly budget counts requests, reviews, automations and Apps charges together, with alerts at 50, 80, 95 and 100 percent.
Do we get invoices?
Every top-up has a receipt, and teams with monthly fees get one statement on the 1st. Larger teams can be invoiced monthly on a credit limit; talk to us.
Is there a free trial?
A new team can claim $5 of credit by verifying a card, which is never charged. What’s unspent after 30 days leaves the balance. The Apps Free plan has no time limit.
Start with $5 of credit on us.
Create the team, add credit once, and every product draws from it inside the same budget.