Unverified50% confidenceBenchmarkExact time
The prefill throughput of this setup increases with prompt length, climbing from approximately 290 tokens/s for short prompts to approximately 1,300 tokens/s, with an average prefill throughput of approximately 900 tokens/s.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/3/2026
First Seen
Valid until: 1/1/2027
Sources
Related Entities
Related Claims
UnverifiedToken billing is calculated based on the total volume of input and output; the longer a conversation, the more each call pays for an increasingly bloated history.72% similarUnverifiedEach retry resends the complete prompt input tokens and is billed in full; a request with 4,000 input tokens retried 3 times will consume 16,000 tokens.70% similarUnverifiedPrompt Caching在提示词前缀完全相同且超过约1024 token阈值时命中，缓存token通常按原价的25%甚至更低计费70% similarUnverified在Pi里说一句你好只占用约1100的上传Token，仅消耗上下文窗口的千分之四；而在Codex里打招呼要消耗18000 Token，约占7%的上下文窗口69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/970256API
curl https://kongchang.com/api/v1/knowledge/claims/970256MCP
get_claim(id=970256)