Skip to main content

Credits & Usage

Credits are part of Hatz AI's consistent, predictable billing model.

Credits and usage

Hatz Credits are the usage units used for AI activity in the Hatz platform. Each tenant has a monthly credit allowance that is shared across the users and AI features available to that tenant. Monthly credit allowances reset each calendar month.

Credit usage reflects the work performed for your request. It can change based on the model selected, the amount of input and output, the files or context included, and the tools or workflow steps used during a request.

What uses credits?

Credit-consuming activity can include:

  • Secure AI chat and assistants

  • Workflow, app, and agent runs

  • Document analysis, file uploads, and generated outputs

  • API requests and tool actions that call AI models or connected services

For apps, completed AI work can consume credits. For workflows and agents, completed AI or tool work can consume credits even if a later step fails or the full run does not complete. If this happens, review the run details to identify which steps completed before the failure.

Why does usage vary?

The same type of task can use different amounts of credits depending on how much work the model needs to do. Common factors include:

  • The AI model selected for the request. Different models can use different credit multipliers, so a larger or more specialized model may use more credits than a smaller model for the same prompt.

  • The length of the prompt, conversation history, uploaded files, or other input context

  • The length and complexity of the answer or generated output

  • Whether the request uses tools, integrations, web actions, file actions, or API calls

  • How many steps, retries, or model calls are required inside a workflow, app, or agent

Use your recent task history to compare usage for similar work.

Do credits reset?

Credits are available at the tenant level and reset each calendar month. Unused monthly credits generally do not roll over unless your agreement says otherwise.

Monthly usage is measured on a calendar-month boundary in Coordinated Universal Time (UTC). Activity is counted toward the calendar month in which it occurs in UTC, so a request made late on the last day of the month in your local time zone may fall into the next month's usage if it happens after midnight UTC. When you reconcile end-of-month usage, compare it using UTC dates so the boundary lines up.

For package-specific allocations or billing details, review the Billing area in your Admin Dashboard or contact your Hatz account team.

Can I estimate credit usage in advance?

You can usually estimate usage directionally, but exact usage depends on the model and request. A short chat with a smaller model will usually use fewer credits than a long workflow that reads files, uses tools, and generates a detailed output with a larger model.

For most users, Auto Mode is the best starting point because it helps route requests to an appropriate model for the task. For a new rollout or high-volume workflow, monitor the first few days of usage and adjust the model, prompt, files, user caps, disabled models, or workflow design if usage is higher than expected.

Where can admins monitor usage?

Admins can review usage from the Admin Dashboard, Tenant Management, billing-period usage, and users and roles views available to their role. These views are intended to help monitor adoption, spot unusual spikes, and plan capacity.

Usage reporting may not be a token-by-token invoice and may not update instantly. Different views may also answer different questions, such as tenant total, billing-period usage, user-level usage, model usage, or run-level history. If usage looks higher than expected, start by reviewing the available usage views for the tenant, users, models, workflows, apps, agents, API activity, files, and date range involved. Look for high-volume users, repeated workflow runs, larger models, long prompts, uploaded files, tool use, retries, or automation that may be running more often than expected.

Current-period usage views refresh on a short cadence rather than in real time. Recent activity typically appears within about 15 minutes, while longer historical date ranges are cached and refresh less often. If you just ran something and do not see it yet, wait a few minutes and refresh before treating it as missing.

If two usage views do not appear to match, confirm that they are using the same tenant, billing period, date range, time zone, feature scope, and user scope so you are comparing the same activity.

What happens if a tenant reaches its allowance?

If a tenant approaches or reaches its monthly credit allowance, admins may see in-product usage indicators and receive email notifications. If the allowance is reached, some credit-consuming actions may be limited until the allowance resets, the package is upgraded, or an account-specific adjustment is approved. See Upgrade Tenant Package for package upgrade steps. The exact experience can vary by feature, model, role, and tenant configuration.

After package credits are used, eligible work can continue in the workspace’s shared Auto pool, equal to 25% of its active package credits. This is one shared pool, not 25% for each user. User/role limits and account eligibility still apply. Extra Usage is a separate budget for additional paid usage. Rate Limiting and Credit Overage explains the transitions and notifications.

Does ADEL Phone Agent use Hatz Credits?

ADEL Phone Agent usage is measured separately in minutes. It is not treated as billable Hatz Credit usage for chat, workflow, app, agent, file, or API activity. For setup details, see AI Phone Agent Settings.

Does the AI Gateway use Hatz Credits?

Not for opted-in tenants. If your MSP has activated the AI Gateway and your organization has opted in, API requests to the Anthropic Messages and OpenAI Responses endpoints are billed in USD instead of Hatz Credits. Chat, agents, Workshop, the Chat Completions API and everything else above still use credits. See AI Gateway: pay for API usage in USD in the MSP / Admins collection.

How can I manage credit usage?

  • Use Auto Mode when available so Hatz can route the request to an appropriate model for the task.

  • Choose the model that matches the task. Use larger models when the task needs them, and smaller models for simpler work.

  • Disable higher-multiplier models for users or groups that do not need them.

  • Use the available model-access controls and custom-role usage limits. Where offered, monthly and weekly limits have independent Auto only or Block usage actions; Auto only keeps consuming credits and is different from the shared Auto pool.

  • Keep prompts and uploaded context focused on the information needed for the task.

  • Review apps that run multiple model calls, and review workflows or agents that run multiple model or tool steps.

  • Use role-based credit limits where you need monthly usage caps for a group of users.

  • Monitor usage regularly after onboarding a new team, workflow, or integration.

When should I contact support?

Most high-usage questions can be narrowed down by reviewing usage reports first. Contact support when the reports do not match a specific request or workflow run, a user or tenant appears to be limited despite available credits, or you cannot reconcile a specific spike after checking users, models, workflows, files, tools, API activity, and date ranges. Include the tenant name, user, date and time, feature used, model selected, and any workflow, file, API, or usage-report details that may help us investigate.

Credit usage support covers usage, limits, resets, unexpected consumption, and blocked-by-limit questions. Billing support covers invoices, payment, refunds, cancellation, downgrades, package changes, procurement, and commercial terms. For those billing or commercial topics, contact the account owner or Billing team rather than relying only on usage reports.

A task can contain several operations

A research-and-report task can search for information and then do further work to prepare the report. Review the credits for the whole task and the useful result, rather than treating one visible message as one model call. Completed work can consume credits even if a later step fails.

An example of the three balances

A package of 10,000 credits has a shared Auto pool of 2,500 credits where continuation applies. Extra Usage is a separate paid-usage budget that an eligible workspace can enable. Review its saved enabled state and budget, and keep independent user/role limits in mind. Custom Roles - Credit Limits explains weekly and monthly controls.

Cedar Project Team role with separate monthly and weekly credit policies.

Extra Usage threshold dialog for Cedar, including billing consent and Cancel.

Watch: combine weekly and monthly credit limits

Watch: understand Auto allowance and Extra Usage

Filter usage for one client

On the Admin Dashboard, select a client and a date range. In the seven-day Cedar Workshop view, Active Users showed four people. Open Breakdown in Credit Usage Trend to see how chats, agents and other features contributed to credit use.

Did this answer your question?