Token Usage

Track how many tokens the agent consumes and what it costs.

/usage command

/usage

Shows token consumption and estimated cost for the current session and totals.

What's tracked

Every API call records:

• Input tokens — prompt + context sent to the model
• Output tokens — model's response
• Model used — which model handled the request
• Cost — calculated from pricing in models.json

Pricing data

Cost is derived from the pricing field in ~/.abtars/config/models.json:

{
  "name": "openrouter/anthropic/claude-sonnet-4",
  "pricing": {
    "input": 3.0,
    "output": 15.0
  }
}

Prices are per million tokens. abTARS multiplies token counts by these rates to estimate cost.