[KongchangAI]
Unverified50% confidenceFactExact time

该1-bit量化的27B模型能够在仅8GB内存的设备上运行

1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/25/2026
First Seen
Valid until: 11/23/2026

Sources

1-bit量化突破：27B大模型跑进8GB内存

Related Entities

Qwen3model

Related Claims

UnverifiedUnsloth推出的1-bit量化版本将27B参数级别的模型压缩至仅6-8GB78% similarVerified一个原本需要60GB显存的30B参数模型，经过量化后能在16GB甚至8GB显存的消费级显卡上运行76% similarUnverified在16GB内存硬件条件下只能容纳5GB以内的模型，因此该开发者使用了体积约4.8GB的Qwen 2.5 7B 4bit量化模型，配8K上下文窗口75% similarUnverified典型的TinyML设备RAM仅有256KB到1MB75% similar

Cite This Claim

Stable URI
https://kongchang.com/claim/798064
API
curl https://kongchang.com/api/v1/knowledge/claims/798064
MCP
get_claim(id=798064)