查询消费用量

明细列表，每条记录对应一次调用：

{
  "success": true,
  "message": "",
  "data": [
    {
      "id": 90210,
      "user_id": 1,
      "created_at": 1763088289,
      "type": 2,
      "content": "model=gpt-image-2.5-flare, quota=45000",
      "username": "alice",
      "token_name": "prod-key",
      "model_name": "gpt-image-2.5-flare",
      "quota": 45000,
      "prompt_tokens": 120,
      "completion_tokens": 0,
      "use_time": 42,
      "is_stream": false,
      "channel": 3,
      "token_id": 12,
      "group": "default",
      "ip": "203.0.113.24",
      "request_id": "20261114123abc"
    }
  ]
}

• data[].quota (integer) — 本次调用的扣费额度。
• data[].model_name (string) — 实际使用的模型。
• data[].prompt_tokens / completion_tokens (integer) — token 用量。
• data[].use_time (integer) — 耗时，秒。
• data[].token_name (string) — 发起调用的 Key 名称。
• data[].created_at (integer) — 发生时间，Unix 秒。
• data[].request_id (string) — 请求 ID，排查时用它对齐上游日志。
• data[].channel (integer) — 渠道 ID。

A detail list where each record is one call:

{
  "success": true,
  "message": "",
  "data": [
    {
      "id": 90210,
      "user_id": 1,
      "created_at": 1763088289,
      "type": 2,
      "content": "model=gpt-image-2.5-flare, quota=45000",
      "username": "alice",
      "token_name": "prod-key",
      "model_name": "gpt-image-2.5-flare",
      "quota": 45000,
      "prompt_tokens": 120,
      "completion_tokens": 0,
      "use_time": 42,
      "is_stream": false,
      "channel": 3,
      "token_id": 12,
      "group": "default",
      "ip": "203.0.113.24",
      "request_id": "20261114123abc"
    }
  ]
}

• data[].quota (integer) — quota charged for this call.
• data[].model_name (string) — the model actually used.
• data[].prompt_tokens / completion_tokens (integer) — token usage.
• data[].use_time (integer) — duration in seconds.
• data[].token_name (string) — the key that made the call.
• data[].created_at (integer) — when it happened, Unix seconds.
• data[].request_id (string) — request ID, useful for matching upstream logs.
• data[].channel (integer) — channel ID.

只想看总消耗与速率时用 GET /api/log/self/stat，参数与明细接口相同，返回一个汇总对象：

{
  "success": true,
  "message": "",
  "data": {
    "quota": 789500,
    "rpm": 12,
    "tpm": 8400
  }
}

• data.quota (integer) — 时间窗口内的总扣费额度。
• data.rpm (integer) — 该窗口内的每分钟请求数。
• data.tpm (integer) — 该窗口内的每分钟 token 数。

For totals and rates only, use GET /api/log/self/stat with the same parameters; it returns one aggregate object:

{
  "success": true,
  "message": "",
  "data": {
    "quota": 789500,
    "rpm": 12,
    "tpm": 8400
  }
}

• data.quota (integer) — total quota charged in the window.
• data.rpm (integer) — requests per minute in the window.
• data.tpm (integer) — tokens per minute in the window.

两个时间戳都用 Unix 秒（10 位）。普通用户的查询范围上限由实例的「最长日志时间范围」配置决定，默认 365 天；超出范围的请求会被截到允许区间。

日志是逐条明细，数据看板是按时间、模型、用户或渠道聚合后的统计。两者的筛选范围与刷新时机可能不同，新请求先出现在日志里，图表要等聚合后才更新。

Both timestamps are Unix seconds (10 digits). The maximum range for regular users comes from the instance's "longest log range" setting, 365 days by default; wider requests are clipped to the allowed window.

Logs are per-call detail, while the dashboard aggregates by time, model, user, or channel. Their filters and refresh timing can differ: a new request shows up in logs first and in charts after aggregation.

更多状态码与重试建议见错误码与限流，常见报错的排查步骤见常见问题。

For more status codes and retry guidance see Error codes and rate limits; for troubleshooting see the FAQ.

按天取消费明细

把 start_timestamp 与 end_timestamp 设成某一天的首尾，再加 type=2，就得到当天的全部扣费记录。

按模型归因

加 model_name 过滤，可以算出单个模型的成本占比。

只取总量

换成 /api/log/self/stat 只返回一个汇总对象，适合做看板上的数字。

Daily usage detail

Set start_timestamp and end_timestamp to the start and end of a day and add type=2 to get every charge that day.

Attribute by model

Add a model_name filter to compute a single model's share of cost.

Totals only

Swap in /api/log/self/stat for one aggregate object, ideal for dashboard figures.