Unverified50% confidenceBenchmarkExact time

Qwen3-VL-8B模型加载就占用约17-18GB显存

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

9/6/2026

First Seen

Valid until: 12/5/2026

Sources

Qwen3-VL本地部署与微调实战：从环境配置到电路板识别

Related Entities

Qwen3-VL-8Bmodel

Related Claims

Unverified以约160GB的模型（如Qwen3.5-397B的UD-Q3_K_XL量化版本）为例，运行透镜需额外约20GB内存78% similar UnverifiedQwen3系列编程模型约20GB加载，推理时可膨胀到35-40GB78% similar UnverifiedQwen3-27B官方发布了GGUF格式的量化模型，最低8GB显存即可完成本地部署76% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/865847

API

curl https://kongchang.com/api/v1/knowledge/claims/865847

MCP

get_claim(id=865847)