Spending Limits

We put no money ceiling on your own provider keys. Managed Keys spends a prepaid balance, so it can run out.

What Managed Keys can run out of

The balance. The money you prepaid. At zero you get 402 insufficient_balance until you add funds.
The daily spending limit. It counts what you spent today plus what your in-flight requests hold, over a UTC day. Past it you get 429 spend_velocity_exceeded until the day rolls over.

A risk review can pause managed requests on top of those two. That’s 403 account_under_review. Your own key routes keep working through all of it.

Two headers report your spend

x-flexinference-spend-remaining says unlimited when no money ceiling applies. Otherwise it’s a signed micro-USD number. It’s the smaller of your balance and what’s left of today’s limit.
x-flexinference-spend-as-of is when we read that number, in epoch milliseconds.

A negative number means we’re turning away managed requests until you add money.

The limits endpoint

Call GET /v1/limits with your key. It lists every ceiling on your organization, spending and usage alike.

curl https://api.flexinference.com/v1/limits \
  -H "Authorization: Bearer $FLEXINFERENCE_API_KEY"

{
  "object": "limits",
  "as_of": 1769558400123,
  "own_keys": {
    "unlimited": true,
    "flex_race_slots": {
      "limit": 40,
      "scope": "organization",
      "on_exceed": "degrade_to_standard"
    }
  },
  "managed": {
    "providers": ["anthropic", "openai"],
    "serving": "ok",
    "paused_reason": null,
    "binding": "daily_spend",
    "spend_remaining_micro_usd": 41200000,
    "currency": "USD",
    "balance": {
      "remaining_micro_usd": 94880000,
      "cash_micro_usd": 84880000,
      "credit_micro_usd": 10000000,
      "reserved_micro_usd": 120000
    },
    "daily_spend": {
      "limit_micro_usd": 250000000,
      "used_micro_usd": 208800000,
      "remaining_micro_usd": 41200000,
      "window": "utc_day",
      "resets_at": 1769644800
    },
    "concurrency": {
      "limit": 10,
      "scope": "organization",
      "on_exceed": "refuse"
    },
    "ramp_week": 1,
    "ramp_defaults": {
      "concurrency": 10,
      "daily_spend_micro_usd": 250000000
    },
    "override_in_force": {
      "concurrency": false,
      "daily_spend": false
    },
    "rates": []
  },
  "abuse_limits": {
    "counted_per_cloudflare_location": true,
    "request_body_bytes": 50000000,
    "rates": [
      {
        "name": "organization_requests",
        "scope": "organization",
        "window_seconds": 60,
        "limit": null,
        "on_exceed": "refuse"
      },
      {
        "name": "api_key_requests",
        "scope": "api_key",
        "window_seconds": 60,
        "limit": null,
        "on_exceed": "refuse"
      },
      {
        "name": "client_ip_requests",
        "scope": "client_ip",
        "window_seconds": 60,
        "limit": null,
        "on_exceed": "refuse"
      },
      {
        "name": "failed_authentications",
        "scope": "client_ip",
        "window_seconds": 60,
        "limit": null,
        "on_exceed": "refuse"
      }
    ]
  }
}

Every money field is a micro-USD number, and one USD is a million of them. as_of is epoch milliseconds.

managed comes back null when no provider runs on Managed Keys.
managed.serving reads ok or paused. When paused, paused_reason carries the matching error code from the errors page.
managed.binding names the ceiling closest to stopping you (balance or daily_spend, with paused and exempt for the special cases).
managed.spend_remaining_micro_usd matches the x-flexinference-spend-remaining header.
managed.balance.remaining_micro_usd adds cash and credit together. reserved_micro_usd is money your in-flight requests hold, and we don’t subtract it.
managed.ramp_week says which week of the account-age schedule you’re in, and ramp_defaults is what that week gives you. override_in_force means an admin set a number by hand, so the schedule won’t move it.
A limit of null in abuse_limits.rates means the number isn’t published. Back off for Retry-After.

When we can’t read one of these records, you get 503 limits_unavailable. Requests keep working.

Get started

Concepts

Additional Features

Using the API

Agent Keys + Integrations

What Managed Keys can run out of

Two headers report your spend

The limits endpoint

​What Managed Keys can run out of

​Two headers report your spend

​The limits endpoint

What Managed Keys can run out of

Two headers report your spend

The limits endpoint