In short. A Brave Answers spending cap is checked before answering, so a running answer that passed the check completes in full.
Definition: No. Brave documents a pre-answer spending check: if the limit has not been exceeded when a question starts, Answers completes that answer even if processing later goes past the limit. The documentation does not give the response returned when a new question starts after the cap.
Key facts
- Brave describes Answers as a usage-based service and says you can set monthly credit limits in your account in its Answers documentation.
- The same documentation says limits are checked before answering, and an answer that starts within the limit is answered in full even if processing exceeds it.
- Brave says the imposed limit caps the amount charged for that answer, according to the Answers documentation.
- For streaming, usage metadata arrives as the last message. For synchronous requests, the same keys appear in response headers.
- Brave's separate rate limiting guide describes a 1-second request window and a 429 response when that request limit is exceeded.
What happens when processing crosses the spending limit?
The spending check happens before the answer starts. Brave's Answers documentation states that if the limit is within bounds when a question starts, the answer is returned in full even when processing later exceeds the limit. Reaching the cap during that running answer therefore does not interrupt the answer under the documented rule. The same sentence says charging stops at the imposed limit.
The rule concerns an answer that has already started. The documentation gives no HTTP status or response schema for a new question that reaches the pre-answer spending check after the cap, so that response cannot be inferred from this page.
How is spending different from request-rate limiting?
Brave documents request-rate limiting as a separate control. Its rate limiting guide says the API uses a 1-second sliding window to count requests for a plan. It also says that exceeding the rate limit returns HTTP 429 and the request fails.
The guide explains that the request count is increased when the API receives the request, while processing time does not affect the rate limit. That timing rule helps distinguish a request-count rejection from the spending rule, which evaluates the question before answering begins.
Which usage signal can a client record?
Brave says each answer includes usage metadata. A streaming client receives that metadata in the last message, while a synchronous client receives the keys in response headers. The documentation's example includes a total-cost field:
<usage>{ "X-Request-Total-Cost": 0.01567 }</usage>
The Answers page also says that single-search mode has lower cost than multiple-search research mode, whose response time can extend to minutes. It recommends checking spending limits before critical operations.
Related terms
FAQ
If the cap is reached while my answer is streaming, does the stream stop?
No. If the question passed the spending check when it started, Brave documents that the answer is completed in full, with charging limited to the imposed cap.
What should I inspect before a critical request?
You should check the spending limit before sending the request. Brave lists that check in its Answers guidance for critical operations.
Does HTTP 429 mean that the spending cap was reached?
No conclusion follows from 429 alone. Brave's rate limiting guide associates 429 with an exceeded request-rate limit, while the Answers page does not publish a status code for its spending-limit check.
Sources
Check how a page reads to answer engines. Open the GEO score analyzer