Skip to content

Token Usage and AI Logs

Ebla AI records aggregate token use and call-level diagnostics.

Token Usage

Open AI Token Usage to review input tokens, output tokens, total tokens, and request counts by user and date.

Token usage dates, request counts, and input and output totals, with user names excluded

Dashboard widgets are available for personal usage and, for authorized administrators, all users.

Effective limit

For a non-admin user, the limit is resolved in this order:

  1. A positive AI Monthly Token Limit on the User
  2. The most permissive AI Token Limit from the user's Teams
  3. Default Token Limit from AI Settings

Despite the user-field label, its value follows the daily, weekly, or monthly period selected in AI Settings.

A Team value of 0 is unlimited. Administrators are not blocked by token limits.

AI Log

Open AI Log for a specific request. Depending on permissions and the request, it can show:

  • Feature and user
  • Provider, resolved model, and profile
  • Start time and duration
  • Input and output tokens
  • Cache status
  • Request and response diagnostics
  • Provider round-trip information
  • Estimated cost when exact model pricing is available
  • Controlled error information

Estimated cost is informational, not billing. It may be empty for unknown or tiered pricing.

AI Log provider, model, token counts, success status, and duration

The request and response view helps compare what was sent with what the provider returned.

AI Log request and response for a fictional account summary

Secrets are redacted. Do not copy API keys into prompts when testing diagnostics.

Troubleshooting a failed request

Record these details before contacting support:

  • Feature
  • Provider and model
  • Approximate request time
  • AI Log ID
  • Whether Test & Fetch Models succeeds

Never send the provider API key.