Unverified95% confidenceFactTime unknown
大模型的API调用通常按Token数量计费，且每个模型都有上下文窗口（Context Window）的Token数量限制
1
Sources
95%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
ReAct模式与Transformer架构详解：前端AI面试高频考点
bilibiliAI全栈进阶Cris
Related Entities
Related Claims
Unverified大语言模型上下文窗口从4K token扩大到百万级token，使长程规划成为可能，模型可在数十步工具调用序列中保持对初始目标的一致理解66% similarUnverified上下文窗口（Context Window）指模型单次交互中能处理的最大token数量65% similarUnverifiedAPI计费同时涵盖输入Token和输出Token，随着对话延伸，每次请求携带的输入Token会线性增长63% similarUnverifiedAPI calls to LLMs are billed by token count, covering both input and output tokens62% similarUnverifiedtoken消耗的精确计量依赖于对Cursor官方API响应体中usage字段（包含prompt_tokens、completion_tokens）的实时解析62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/22370API
curl https://kongchang.com/api/v1/knowledge/claims/22370MCP
get_claim(id=22370)