Unverified50% confidenceTradeoffExact time
Using lightweight embedding models such as sentence-transformers' MiniLM requires tens of milliseconds per inference, while large embedding models can reach hundreds of milliseconds.
1
Sources
50%
Confidence
Long-term
Relevance
9/9/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedA single LLM inference call can easily take hundreds of milliseconds or even seconds, whereas the execution time of deterministic code is typically at the microsecond level.77% similarUnverified对于BERT-base模型推理会话初始化通常耗时数百毫秒，重复执行会使延迟增加1-2个数量级70% similarUnverifiedRule- and classifier-based routing solutions are typically built on DistilBERT, sentence embedding models, or fine-tuned small language models, with inference latency generally kept within 10 milliseconds.67% similarUnverifiedIn actual testing, a single classification by the Dify question classifier takes approximately 11 seconds, because execution speed depends on the large model's response.66% similarUnverified用于文本分类的小型 BERT 变体参数量约 1 亿，单次推理延迟通常在毫秒级66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/881438API
curl https://kongchang.com/api/v1/knowledge/claims/881438MCP
get_claim(id=881438)