Unverified50% confidenceFactExact time
Different models price input and output tokens differently, and some providers have differentiated pricing for scenarios such as cache hits and batch requests.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Building Your Own LLM Proxy: The Insider Story of Model Calls You Didn't Know
hackernewshackernews10/18/2026
Related Entities
Related Claims
UnverifiedThe unit price of input tokens is usually cheaper than that of output tokens, but input tokens are billed on every request.74% similarUnverifiedUsage-based billing is the opposite of the quota system: users do not pay for a fixed allowance, but are settled according to the actual number of tokens or requests called, with the price growing linearly with usage.74% similarUnverifiedRequiring the model to output short answers or a fixed format often saves more cost than trimming the prompt, because output is usually the more expensive side.73% similarUnverifiedSubscription-based quotas are usually not priced directly in dollars, but instead use credits or abstract quota units that are converted internally by the platform according to model weights, creating information asymmetry72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1069639API
curl https://kongchang.com/api/v1/knowledge/claims/1069639MCP
get_claim(id=1069639)