Your limits

One call returns everything that can block a request on your account, so a dashboard can show quota bars and remaining daily budget without guessing.

Response

{
  "user_id": "612aabd4-...",
  "balance_micro": 9950954,
  "balance_usd": 9.950954,
  "token_limit": 0,
  "tokens_used": 122008,
  "tokens_remaining": -1,
  "token_multiplier": 1.0,
  "allowed_models": [],
  "price_overrides": {},
  "network_fee_micro": -1,
  "daily_model_caps": [{
    "model": "deepseek-v4.1-flash-1017:free",
    "name": "DeepSeek V4.1 Flash 1017 (free)",
    "input":  { "limit": 700000, "used": 82,    "remaining": 699918 },
    "output": { "limit": 300000, "used": 32905, "remaining": 267095 },
    "resets_at": 1792540800
  }]
}

• tokens_remaining is -1 when the account quota is unlimited.
• remaining is -1 for a side with no cap.
• resets_at is the next UTC midnight (unix seconds).
• daily_model_caps lists only models that have caps; an empty array means no per-model caps apply.

Per-model daily caps

A model row can carry daily_input_tokens / daily_output_tokens (0 = unlimited). They are enforced per account, across all keys, from UTC midnight to UTC midnight.

• Input: a request whose prompt would push today's input total over the cap is rejected with 403 before the upstream call.
• Output: once the cap is reached the request is rejected; otherwise max_tokens is clamped to the remaining allowance.
• Accounting uses provider-reported tokens: input includes cached input, output includes reasoning.
• The account owner gets an email at 80% and 100% of either cap (once per threshold per day).

Account quota

token_limit caps total tokens across the account's lifetime. When it is reached, requests fail with 403 account token quota exhausted. token_multiplier scales how fast the counter grows; it does not change the amount billed.

Model access

allowed_models lists the models this account may call. An empty array means every public model. A non-empty key-level allowlist still wins over this list.

price_overrides replaces catalog prices at billing time in the form {"model-id":[prompt_per_m, completion_per_m]}. network_fee_micro overrides the per-request fee: -1 uses the model default, 0 waives it.