Skip to main content

Understand session usage

Use session usage when a conversation is becoming long, a model feels expensive, or you want to understand how much context OpenCode is carrying.

Open the usage sheet​

Tap the usage indicator in the Chat header. The sheet summarizes the current session and active model.

Session usage showing context utilization and token activity

Read the numbers​

  • Context utilization is the portion of the model's input window currently represented by session input.
  • Input and output tokens measure text sent to and returned by the model.
  • Reasoning tokens are reported separately when the provider supplies them.
  • Cache read/write shows reused or newly cached context when supported.
  • Steps counts recorded assistant processing steps.
  • Estimated cost appears only when the server has usable pricing metadata.

“Pricing unavailable” does not mean the request was free. It means the app cannot calculate a reliable amount. “Included or unpriced” may represent subscription-included usage or missing provider pricing.

Use the information​

  • Start a fresh session when unrelated history is consuming context.
  • Keep one session focused on one outcome.
  • Choose a smaller model for routine follow-ups when appropriate.
  • Treat cost as an estimate and use the provider's billing dashboard as the source of truth.

Next: Manage sessions, configure models, or start a focused task.