Skip to content

Public Stats

Read anonymized, aggregated public usage statistics. No authentication required.

All endpoints below are public — no auth is required and only anonymized aggregates are returned; no personal information or individual request content is included. The per-app token leaderboard (GET /v1/stats/apps) is covered on the App Attribution page.

Per-model call stats#

GET/v1/stats/models

Returns the top 20 models by call count in the period, plus the total call count. The default period is daily.

ParameterTypeRequiredDescription
periodstringOptionaldaily | weekly | monthly.
200 OK
{
  "period": "daily",
  "generated_at": "2026-07-03T09:00:00+00:00",
  "total_calls": 4210,
  "top_models": [
    {"model": "gpt-4o-mini", "calls": 1840},
    {"model": "claude-sonnet-4-6", "calls": 920}
  ]
}

Leaderboard & market share#

GET/v1/stats/rankings

The counterpart of OpenRouter's rankings. Returns the top-20 model leaderboard (leaderboard) ranked by token usage and the top-12 provider market share (market_share). Failed calls (status code 400 or above) are excluded. The default period is weekly.

ParameterTypeRequiredDescription
periodstringOptionaldaily | weekly | monthly.
200 OK
{
  "period": "weekly",
  "generated_at": "2026-07-03T09:00:00+00:00",
  "total_tokens": 48210000,
  "leaderboard": [
    {
      "rank": 1,
      "model": "claude-sonnet-4-6",
      "provider": "anthropic",
      "maker": "Anthropic",
      "tokens": 21400000,
      "calls": 8120,
      "trend_pct": 12.4
    }
  ],
  "market_share": [
    {"provider": "anthropic", "tokens": 21400000, "share_pct": 44.4},
    {"provider": "openai", "tokens": 15200000, "share_pct": 31.5}
  ]
}

maker is the model's creator (lab) while provider is the hosting provider. trend_pct is the token change (%) versus the previous period, and is null when there is no prior data. share_pct is the share (%) of the top-12 provider total.

Token usage timeseries#

GET/v1/stats/usage-timeseries

A per-bucket × top-N-model token usage timeseries. The top N models by total tokens over the range get individual series; the rest are folded into "Others". Only the token-billed (text) modality is aggregated, and only token counts are returned — cost is not exposed.

ParameterTypeRequiredDescription
rangestringOptional30d | 12w | 24w. Default 30d.
granularitystringOptionalday | week. Default day. Weeks are ISO weeks (Monday start).
topintegerOptionalNumber of top models exposed as individual series. 1 to 20, default 8.
200 OK
{
  "range": "30d",
  "granularity": "day",
  "generated_at": "2026-07-03T09:00:00+00:00",
  "series": ["gpt-4o-mini", "claude-sonnet-4-6", "Others"],
  "points": [
    {
      "t": "2026-07-02",
      "tokens": {"gpt-4o-mini": 820000, "claude-sonnet-4-6": 610000, "Others": 90000},
      "total": 1520000
    }
  ],
  "total_tokens": 48210000
}
Early on with low traffic, lists may be empty and trend_pct may be null.