Unverified50% confidenceFactExact time
In Agent paradigms such as ReAct, the model must process the complete history of all previous steps at each step, causing the cost impact of payload size to grow quadratically rather than linearly.
1
Sources
50%
Confidence
Long-term
Relevance
8/8/2026
First Seen
Sources
13款搜索API实测：你忽略的隐藏成本可能比标价还高
Related Claims
Unverified在ReAct框架下，每次工具调用需将完整历史上下文重新传入模型，导致上下文窗口随执行步骤线性累积，而计费成本呈近似二次方增长79% similarUnverified自回归生成机制决定了推理延迟与成本随输出长度线性增长73% similarUnverifiedThe cost of each large language model call is positively correlated with the number of input tokens, and in an Agent loop, as the conversation history accumulates, the cost of each call grows linearly.70% similarUnverifiedStrategies for cost and latency optimization include result caching, parallel execution of tasks with no dependencies, and model tiering.70% similarUnverifiedBefore ReAct was proposed, large models' reasoning capabilities (represented by Chain-of-Thought) and action capabilities (represented by Action Generation) were two independent lines of research69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/711416API
curl https://kongchang.com/api/v1/knowledge/claims/711416MCP
get_claim(id=711416)