Usage and billing
Aleph attributes runtime usage to the account that owns the agent while retaining the agent ID on every event.
Billing accounts
| Agent placement | Billing account |
|---|---|
| Personal | Creating user |
| Organization | Organization |
| Team | Organization, with team attribution |
This is similar to repository hosting: an organization pays for shared work, while teams organize access and reporting rather than forming separate billing accounts.
Usage credits
Each account has one monthly usage-credit balance. The Usage page keeps the categories visible so you can understand what consumed it:
- model usage, based on the input and output tokens and model used;
- isolated runtime, based on the resources used while an agent works;
- web searches, based on each search request (at most two web searches per agent turn);
- X searches, based on each live X (Twitter) post search request (at most one X search per agent turn);
- connected-app tool calls, when an agent uses authorized Connections tools;
- turns, which attribute chat, schedule, and channel activity without charging for the same work twice.
Free includes 250 credits each month. Pro is $29 per month and includes 8,000 credits. Max is $100 per month and includes 30,000 credits. Credits reset at the start of each UTC calendar month and do not roll over.
Aleph warns at 80% of the account allowance and stops new agent runs at 100%. There are no automatic overage charges. Per-agent attribution explains where consumption came from, while personal or organization Billing shows the shared remaining balance.
Platform model access is included in this accounting model; there is no user-supplied model key today. Checkout and invoice availability can depend on deployment configuration, while usage tracking itself does not. Every plan can use Aleph Light or Aleph Pro. Light prioritizes fast, efficient work; Pro is intended for deeper reasoning and complex multi-step tasks. The selected model's measured token cost is converted into the same usage credits. Connected-app events have a separate monthly allowance because a noisy source can start many runs even when each event is inexpensive. Free includes 500 accepted trigger events and 3 active trigger subscriptions; Pro includes 10,000 events and 25 active subscriptions; Max includes 50,000 events and 100 active subscriptions. Only the first accepted delivery of a provider event counts. Provider retries, Aleph delivery retries, and synthetic tests do not consume this allowance. The resulting agent turn, model usage, sandbox time, searches, and approved tools continue to use their normal meters.