Spend cap

The spend cap is a hard ceiling on month-to-date (MTD) spend, enforced on every chat call. You cannot accidentally overshoot it from concurrent traffic, and you cannot disable it. Picking a cap is part of onboarding.

Cap basics

• Range: $1 — $10,000 per month.
• Set during onboarding; editable any time from the console.
• Resets at the start of each billing period.
• Free credit (default $25) burns down before paid spend. The cap counts total committed cost — credit + paid — so a $5 cap means $5 total this period including credit.

What gets enforced per call

Each chat call:

1. Is rejected with 402 onboarding_incomplete if you haven't picked a cap yet.
2. Is rejected with 402 and a specific reason if your account is currently blocked (spend_cap_exceeded, payment_failed, subscription_canceled).
3. Otherwise, an upper-bound cost for the call is checked against your remaining headroom. If the call would push you past the cap, you get 402 spend_cap_exceeded and the call never runs.

The upper bound is calculated from the input length plus the maximum output size you've requested. Two practical implications:

• Output size is capped at 4,096 tokens per call. A larger max_tokens in your request is silently clamped down.
• Input above 32,768 tokens rejects with 413 input_too_large before any spend is reserved.

After the call completes, the actual cost (which is always lower than the reservation) is settled against your MTD spend.

Mid-stream cap abort

For streaming completions, we re-check your MTD against the cap during the response. If you cross the cap mid-stream, the SSE stream emits an error event and closes:

event: error
data: {"code":"spend_cap_exceeded","message":"Spend cap reached during this request. Raise your cap to continue.","docs_url":"https://docs.olava.dev/billing/spend_cap_exceeded"}

Chunks emitted before the abort are billable. Subsequent chat calls return 402 spend_cap_exceeded immediately.

What 402 spend_cap_exceeded looks like

{
  "error": {
    "type": "billing_error",
    "code": "spend_cap_exceeded",
    "reason": "spend_cap_exceeded",
    "message": "Your monthly spend cap has been reached. Raise your cap to continue.",
    "docs_url": "https://docs.olava.dev/billing/spend_cap_exceeded",
    "resolution_url": "https://console.olava.dev/billing",
    "portal_url": "https://console.olava.dev/billing"
  }
}

Response headers include Retry-After, seconds until your current billing period ends. Well-behaved clients should back off until then or until you raise the cap.

Recovering

Two ways out:

1. Raise the cap from the console. When the new cap is above your committed MTD, the block is cleared immediately so the next chat call goes through.
2. Wait for the period rollover. Caps reset at the start of each billing period; Retry-After tells you when.

Editing your cap

You edit the cap from console.olava.dev/billing. A few rules:

• You can change the cap up to 10 times per 24 hours.
• For large raises (more than 2× the current cap, or more than $1,000), the console will ask you to sign in again before the change takes effect.
• Lowering the cap below your already-committed MTD blocks new chat calls immediately.

Cap warning emails

You'll get warning emails at the 50%, 80%, and 95% thresholds of your cap, plus a notification the moment service is paused for cap reasons. Raising the cap also triggers a confirmation email; lowering does not.