Agent Usage page

The Agent Usage page in the Admin Console provides analytics on how your organization is using Tabnine Agent over a selected timeframe. It helps admins understand adoption, monitor cost, and optimize configuration.

Timeframe

At the top of the page, you can select a monthly timeframe (e.g., “Feb 2026”) from the dropdown menu. You will see a Month trend summary bar that shows:

• Total usage against your monthly budget or quota

  • Percentage used

  • Remaining budget/credits

  • Estimated number of days remaining at the current burn rate

Usage by Model

The Usage by Model table shows how all the available AI models are being used in the above-selected timeframe, according to the following metrics:

• Model name: The name of the AI model. Each row represents one model.

• Total tokens: The total number of tokens processed by that model in the selected timeframe, including both:

  • Input tokens: The number of tokens sent to the model (prompts, instructions, context)

  • Output tokens: The number of tokens generated by the model in responses

• Cache read: The number of tokens served from cache rather than recomputed, if caching is in use

• Cache write: The number of tokens written into cache for potential reuse

• USD: The total cost attributable to this model in the selected timeframe, in US dollars

Usage by Team

The Usage by Team table shows AI usage and costs are distributed across all of your organization’s teams (by the above-selected timeframe).

You will see a breakdown of 1) the percentage of usage conducted by team, and 2) the cost of each team’s operations.

Usage by Tool/MCP

The Usage by Tool/MCP table shows the most common tools (native tools and commands) and MCP servers by users in the organization, organized by the number of calls.

Agent Analytics

The Agent Analytics charts show progress over a given timeframe (default 12 months, but you can change this to the right above the charts).

The four charts monitor the:

1. Unique Users: Number of users using the Agent

2. Agent Messages: Number of Agent messages

3. Messages per User: Number of Agent messages per user

4. Lines of Code: Number of lines of code generated