Tokens and cost
Note: Costs depend on the model name PraisonAI sees for each call. Unknown models show as unknown, not a guessed price.
What you are paying for
Each model call sends input tokens (your prompt and context) and returns output tokens (the reply). Cached tokens, when the provider reports them, can lower the input side.
After a praisonai run with a session, the CLI can print a short footer with token counts and an estimated cost.
Example
praisonai run "Summarise this in one line: agents use tools" --session observe-demoExample explained
- input tokens — What you sent to the model.
- output tokens — What the model wrote back.
- –session — Names the run so totals can accumulate if you continue later.
See totals across sessions
Session list shows stored token and cost columns when runs used a session id.
Usage rolls up many sessions when you want a weekly view.
Example
praisonai session list
praisonai usage --days 7Reference: documentation