Usage and models

Where cost appears

Token totals and estimated cost may show on the Usage page and sometimes on the active session in Chat.

Those numbers come from the model ID configured on the agent, not from a per-message picker in the composer.

Change the model

Set llm on the Agent in Python, or model in agents.yaml, then restart or reload the console if your build caches agent config.

Use praisonai models validate on the CLI before you paste a new ID into YAML.

Example

agents:
  - name: Helper
    instructions: Answer briefly.
    model: gpt-4o-mini

Example explained

• model — The ID the runtime sends to the provider for this agent.

CLI totals still apply

The same session ids and usage commands from the Observe course work when runs went through the console or the CLI.

Example

praisonai session list
praisonai usage --days 7