DOCUMENTATION

Usage and limits

Monthly token allowances, storage, and where to see what your agent has used.

Model tokens

Standard and Smart include 130M model tokens per agent each calendar month (UTC). Every step the agent takes sends its conversation so far to the model, so most tokens are input it has already read (cached input). Output is a small share.

• We email you when an agent passes 80% of its allowance, and again when it reaches it.
• At the allowance, model work pauses until the 1st of the next month. Files and workspace connections stay as they are.
• You’re never charged for extra tokens unless you turn them on for an agent.
• Bring your own model agents aren’t capped by Acenta; your provider bills usage.

Seeing usage

The agent’s page shows tokens used this month, a 30-day daily chart, and each model request from the last 7 days with its new input, cached input and output.

Storage

Each agent has 10 GB of persistent storage for its files. The agent’s page shows how much is used. Files are kept when the agent stops.

Limits

• Up to 3 workspaces per agent.
• Up to 3 agents created from inside one workspace.