Unverified50% confidenceSolutionExact time
The token count recorded by the Cost node comes from the usage field of the API response; if the provider does not return usage (such as some proxy endpoints), then in_tokens: null is written and usd_estimated is marked, and estimated rows do not enter alert queries but only enter the dashboard.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedOpenAI reports cached input tokens in usage.prompt_tokens_details.cached_tokens, and the billing rate is lower; ignoring this causes costs to be overestimated.72% similarUnverifiedThe additional token cost consumed by running the safety classifier is borne by Anthropic and is not passed on to users.69% similarUnverifiedIn mainstream API pricing structures such as Anthropic's, the unit price of cache-read tokens is only 1/10 to 1/30 of standard input tokens.69% similarUnverifiedThe Server-side Classification of certain Agents consumes additional Tokens, and can be skipped at startup to reduce overhead.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1032093API
curl https://kongchang.com/api/v1/knowledge/claims/1032093MCP
get_claim(id=1032093)