Skip to content
LLMTR Docs

Usage Tracking

After each API request, the following fields are recorded:

  • input_tokens — prompt token count
  • output_tokens — completion token count
  • total_tokens — sum
  • latency_ms — provider + gateway total latency
  • status_code — HTTP status
  • cost_usd — spend in USD
  • model — canonical ID
  • api_key_prefix — first 12 characters of the key

Dashboard filters

Section titled “Dashboard filters”

Dashboard > Usage supports:

  • Date range (today, 7d, 30d, this month, last month)
  • Model filter
  • API key filter
  • Status code filter (success / error)

CSV export

Section titled “CSV export”

Click Download CSV in the top right. The same data source feeds the dashboard and the export — consistency guaranteed.

Programmatic access

Section titled “Programmatic access”
Terminal window
curl "https://llmtr.com/api/usage?from=2026-04-01&to=2026-04-30" \
  -H "Authorization: Bearer sk_your_key"

Response:

{
  "data": [
    {
      "timestamp": "2026-04-18T12:34:56Z",
      "model": "openai/gpt-4o",
      "input_tokens": 124,
      "output_tokens": 587,
      "total_tokens": 711,
      "cost_usd": 0.00356,
      "latency_ms": 1240,
      "status_code": 200,
      "api_key_prefix": "sk_abc..."
    }
  ],
  "meta": {
    "page": 1,
    "perPage": 25,
    "total": 142
  }
}

Aggregate metrics

Section titled “Aggregate metrics”

The overview dashboard shows:

  • Total tokens (last 30 days)
  • Total spend (USD)
  • Request count
  • Average latency
  • Most-used model