Unverified50% confidenceFactExact time
In the caching mechanism, output tokens do not benefit from the input cache discount, causing Claude samples to gain less from caching because more of the API value is spent on output tokens.
1
Sources
50%
Confidence
Long-term
Relevance
9/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedClaude's input tokens are typically approximately 3–5 times cheaper than output tokens, because the generation process requires autoregressive token-by-token inference.72% similarVerifiedIn API billing, the unit price of output tokens is generally higher than that of input tokens, because the generation process requires autoregressive inference token by token70% similarUnverifiedAnthropic's Claude API offers a Prompt Caching feature that reduces the input price by 90% on cache hits.70% similarUnverifiedCached input tokens typically enjoy a discount of 25%–50% off the original price, as there is no need to recompute embeddings and attention weights—only the precomputed KV vectors need to be loaded from cache.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/924289API
curl https://kongchang.com/api/v1/knowledge/claims/924289MCP
get_claim(id=924289)