Usage and quota

Reports what your key has spent against the shared model APIs, and what’s left today.

GET /api/profile/model-usage Session or API key

Request

Parameter	Type		Description
include_recent	boolean	Optional	Include a list of recent individual calls. Defaults to false.

curl "https://radeon-global.anruicloud.com/api/profile/model-usage?include_recent=true" \
  -H "Authorization: Bearer $RADEON_API_KEY"

Response

{
  "status": "ok",
  "rpm_limit": 30,
  "daily_cost_limit_usd": 10,
  "daily_cost_used_usd": 1.8412,
  "daily_cost_remaining_usd": 8.1588,
  "daily_reset_timezone": "Asia/Shanghai",
  "daily_reset_at": "2026-08-26T00:00:00+08:00",
  "today": {
    "requests": 142,
    "errors": 3,
    "total_tokens": 88214,
    "prompt_tokens": 31002,
    "completion_tokens": 57212,
    "cost": 1.8412,
    "last_request_at": "2026-08-25T14:22:07+08:00"
  },
  "last_30_days": { "requests": 2840, "total_tokens": 1904221, "cost": 39.77 },
  "all_time":     { "requests": 9120, "total_tokens": 6210443, "cost": 128.44 },
  "by_model": [
    { "model": "DeepSeek-V4-Flash", "requests": 120, "total_tokens": 74001, "cost": 1.55 }
  ],
  "recent": []
}

Fields

Field	Description
status	ok normally. not_configured if the account has no API key yet; not_available if the usage service can’t be reached.
rpm_limit	Requests per minute allowed for this key.
daily_cost_limit_usd	Daily spend ceiling.
daily_cost_used_usd	Spent so far in the current window.
daily_cost_remaining_usd	What’s left before requests start being rejected.
daily_reset_timezone	Timezone the daily window is anchored to — Asia/Shanghai.
daily_reset_at	When the current window rolls over.
today, last_30_days, all_time	Aggregates over each period.
by_model	The same aggregates broken down per model.
recent	Individual recent calls. Empty unless include_recent=true.

Token counts split into prompt_tokens and completion_tokens; some models also report reasoning_tokens. Costs are USD.

Quota values are per account

The numbers above are an example. Your actual limits are whatever this endpoint returns for your key — read them rather than assuming, since they differ between accounts and change over time.

Using it

Poll this before a large batch job to confirm there’s headroom, or watch daily_cost_remaining_usd in a dashboard so a job doesn’t stop halfway through the night. Once the daily allowance is exhausted, requests are rejected until daily_reset_at.

Credits for GPU instances are tracked separately and don’t appear here — see Account.

Errors

502 means the usage service is unreachable; the aggregate fields come back zeroed with a status explaining why. Treat a 502 as “unknown”, not as “zero usage”.