Unverified50% confidenceDecision RuleExact time

端侧推理中吞吐量需达到5-10 token/s对应人类舒适阅读速度，低于此值输出显得一字一顿

1

Sources

50%

Confidence

Long-term

Relevance

7/17/2026

First Seen

Sources

Bonsai 27B: The First 1-bit LLM That Runs on Your Phone — A New Milestone for On-Device AI

Related Claims

Unverified高推理强度模式下单次请求的 Token 消耗可能是低强度模式的 5-20 倍71% similar UnverifiedResearch indicates that when the context exceeds a certain length, the model's recall accuracy for content in the middle portion may drop below 50%.66% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/544953

API

curl https://kongchang.com/api/v1/knowledge/claims/544953

MCP

get_claim(id=544953)