Skip to main content
Billing settings show the workspace plan, usage, limits, and billing portal entry points available to workspace owners and admins.

What admins can manage

Use the billing area to review:
  • Current plan.
  • Usage by product category.
  • Limits for API, logs, evaluations, Behaviors, and Improve cycles when they apply.
  • Billing portal access.
  • Trial, upgrade, or support actions shown for the workspace.

Plan limits

Plan limits keep usage predictable across teams and environments. When a limit is close, decide whether to reduce traffic, archive unnecessary test data, adjust sampling, or contact Adaline for a plan change.

Usage categories

The categories shown in billing can map to different parts of Platform:
CategoryRelated product area
SeatsWorkspace members
Projects per teamspaceTeamspace organization
Prompts per projectPrompts library
Datasets per projectDatasets library
Rows and columns per datasetDataset design
Evaluators per promptPrompt evaluation workflow
Deployment environmentsDeploy
Deployment readsRuntime handoff and SDK/API usage
Log writes and readsMonitor, Traces, Behaviors
Dataset reads and writesEvaluations, API, and automation
Evaluation executionsEvaluators and Datasets
Improve usageGenerated datasets, generated evaluators, and Improve cycles

Admin checklist

  • Keep production API keys named and owned.
  • Review usage after major releases or load tests.
  • Make sure provider credentials are stored only in workspace settings.
  • Restrict billing access to workspace owners and admins.
  • Use View API usage before rotating or deleting keys.

Cost-control habits

  • Sample continuous evaluators when full traffic scoring is not needed.
  • Keep regression datasets curated instead of endlessly growing.
  • Use lower-cost models for evaluator drafts when acceptable.
  • Watch dynamic dataset columns because API and prompt-backed columns add runtime work.
  • Cache deployment reads according to your application architecture.
  • Review Monitor after prompt changes that affect token count, tool calls, or model choice.