Credit and Billable Items¶
Billing Center uses Credit to track your account balance and platform usage. The console displays payment amount, purchased Credit, usage, unit price, and consumption separately—do not conflate these fields when reconciling charges.
Amounts and quantities¶
Field |
Purpose |
|---|---|
Credit balance |
Credit currently available to your account |
Usage |
Quantity consumed under a billable item |
Unit price |
Rate used to calculate Credit consumed by that usage |
Consumption |
Actual Credit deducted for that usage record |
Payment amount |
Currency amount submitted to the payment channel for a top-up |
Purchased Credit |
Credit credited to your account after a successful top-up |
Estimated available days on Cost Overview is an estimate derived from recent consumption rates. Always rely on Usage Details for verified charges; the estimate is not a service-duration guarantee.
Billable items shown in the UI¶
Usage Details opens in the By billable item view. Billable categories include:
Item |
Unit reference |
Scope and how to inspect |
|---|---|---|
Genesis |
Tokens, pages, images, seconds, or minutes depending on model type |
External model calls and inference usage |
AI Service |
Per million tokens or scenario units |
Expand a row to view agent chat, KB Q&A / RAG, reranking, and system-operator image/text embedding |
Compute |
Credit / hour (metered per minute) |
Expand a row to view query/task compute instances and managed runtime environments |
Data Storage |
Credit / GB·month |
Catalog data object storage consumption in deployments with storage metering enabled |
Displayed sub-items depend on active workloads in the selected range. If older labels appear on historical records, MOI AI Service corresponds to AI Service, MOI Compute Hour to Compute, and TaaS Token to Genesis. Verify all charges using the items and records in Usage Details.
How compute instances and sandboxes are billed¶
Compute instances and runtime environments (sandboxes) are metered strictly by active ready occupation—pending startup, failed states, and completed stops do not generate running charges:
1. Compute instances (query and task)¶
Lifecycle state |
Billed? |
Metering and deduction behavior |
|---|---|---|
Starting / Provisioning |
No |
Worker nodes are not ready yet and produce no usage |
Running / Ready |
Yes |
Once nodes reach Ready, exact active replica seconds are recorded in 1-minute (UTC) windows until Stopped, converted to Credit at |
Stopping / Suspended (0 nodes) |
No |
Applies to manual Suspend and idle Auto-suspend scaling down to 0 nodes; running charges stop once nodes terminate |
Failed / Unhealthy |
No |
Failed or unready instances do not generate running minute bills |
Pre-activation balance check: Before activating compute workloads or additional worker types, the system verifies sufficient Credit. Insufficient Credit blocks activation.
Cost control: Enable idle auto-suspend for intermittent workloads, and reserve always-on
minimum = 1configurations for latency-sensitive endpoints. See Compute instances.
2. Agent runtime environments and sandboxes¶
MOI separates runtime environments into ephemeral conversation sandboxes and managed runtime environments, with distinct billing behavior:
Ephemeral conversation sandboxes (not billed separately): Short-lived sandboxes spun up or checked out from a warm pool during ordinary agent conversations do not create compute-binding usage and incur no separate sandbox charge.
Managed runtime environments (billed continuously while ready and bound): A managed environment reserves a workspace compute binding and occupies compute capacity continuously according to its lifecycle phase:
Managed sandbox phase |
Observable status |
Billed? |
Notes |
|---|---|---|---|
Queued after creation |
Waiting for resources |
No |
Reserved while queued for capacity; metering has not started |
Container initialization |
Provisioning (not ready) |
No |
Container image pull and startup do not generate billing |
Ready and retained |
Running / Ready |
Yes |
Metered per minute starting when the environment reaches Ready; even when conversations or terminals are idle, continuous charges accrue until deletion completes |
Deletion to full termination |
Stopping / Deleting → Stopped |
Ends upon full termination |
Charges stop once the binding is fully closed and deleted |
Cost tip: Delete unused managed runtime environments from Resource Center → Runtime Environments and confirm they are stopped/removed so idle reserved bindings do not accrue continuous charges. See Create a managed runtime environment and Terminals, networking, and maintenance.
3. Sample data exemption¶
Workspaces designated for built-in sample-source onboarding exemptions skip compute usage deductions automatically; regular workspaces follow standard metering rules.
Confirm prices¶
Available top-up tiers, currency, exchange rate, and single-order minimum/maximum limits are returned dynamically by the billing service and displayed inside the Top-up dialog. Billing Center does not currently provide a standalone online pricing calculator page; refer to the official pricing documentation, Top-up dialog, rated usage detail unit prices, or your commercial contract for list prices, discounts, validity periods, and enterprise rates, and always distinguish payment amount from purchased Credit.
Relation to MatrixOne instance bills¶
The Credit ledger covers AI Studio (AI Service, Compute, Managed Runtime Environments, Data Storage) and Genesis model consumption. MatrixOne database instance CU, instance storage, public network egress, renewals, and scaling bills are managed separately—see Instance Billing.