Skip to content
AI Radar by Wagner Solution
CS
Menu
context Tools and apps

Anthropic reinstates billing for refused requests in the Claude API for selected safety categories

clearly official source

Anthropic is once again billing for refused requests in the Claude API in the bio, frontier_llm, and reasoning_extraction categories when the refusal occurs before any output – it is billed at the same rate as a regular request.

Anthropic is reinstating billing for refusals in the Claude API in cases where the model refuses a query before generating any output and the refusal falls into the bio, frontier_llm, or reasoning_extraction category. According to the company, these are precisely the categories in which a low rate of false-positive refusals is measured. Such requests are now billed at the same rate as a regular request for the given model.

Refusals that occur in the middle of generating a response (mid-stream) were, according to the company, already being billed before – the change concerns only refusals that occur before any output. Refusals in other safety categories that occur before output remain free of charge, and the so-called fallback credit also remains unchanged. The change applies across all platforms on which the Claude API is available.

What changed

Why it matters

The change directly affects the costs of companies whose products process queries close to the boundaries of safety policy via the Claude API – even a refused response in these categories now counts toward API spending just like a fulfilled request. Companies therefore need to adjust their cost estimates and monitor how often their traffic falls into the bio, frontier_llm, or reasoning_extraction categories.

Two audiences, two different impacts

What this means

01

For individuals

Developers who test or target edge-case queries in the bio, frontier_llm, or reasoning_extraction areas via the Claude API may now pay for a request that the model refuses right from the start, without receiving any output at all.

What to do When testing edge-case queries in the categories listed above, account for the fact that even a refused response will be billed.
More practical updates →
02

For a business

Companies running products on the Claude API must now account for costs even for requests that the model refuses to process in the bio, frontier_llm, or reasoning_extraction categories – these are billed at the same rate as a regular request, which affects the API budget and needs to be factored into cost monitoring.

Development
What to decide Check the volume of requests in the bio, frontier_llm, and reasoning_extraction categories and factor possible refused requests into the estimate of monthly costs for the Claude API.
More business impacts →
Anthropic security billing Claude API refusals

Check the original

Event sources

clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.

1
Claude Platform Release Notes primary source · first detected We're resuming billing for refusals that arrive before any output when stop_details.category is "bio" , "frontier_llm" , or "reasoning_extraction" , the categories where we measure…