Unverified50% confidenceFactExact time
LLM本质上是无状态的推理引擎，每次调用模型时所有记忆都必须以文本形式塞入上下文窗口
1
Sources
50%
Confidence
Long-term
Relevance
7/17/2026
First Seen
Sources
Related Claims
Partially VerifiedEach LLM inference is stateless and constrained by the limited context window; the model has no persistent memory of its own.81% similarUnverifiedThe memory of LLMs depends on the context window, which is the maximum number of tokens the model can process in a single inference.80% similarUnverified大语言模型（LLM）在架构层面是无状态系统，每次推理调用是一次独立的前向传播计算，模型权重在推理阶段完全冻结78% similarUnverifiedLLM本质上是无状态的文本预测系统，没有感知当前运行环境的原生能力78% similarUnverifiedLLM 本质上是自回归文本生成模型，并非专为长程任务规划设计78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/541355API
curl https://kongchang.com/api/v1/knowledge/claims/541355MCP
get_claim(id=541355)