Local Model Hub

Local Model Hub · GGUF

20

Models

3

Category

Q4

Quantization

Clipboard App Companion · CN: hf-mirror (no VPN) / Global: Hugging Face

中 EN

Qwen2.5-1.5B-Instruct

Light daily · clipboard classify · short writing

AlibabaCN

Size1.07 GB

Min RAM8 GB

VRAM2 GB opt

CPU SpeedFast

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

Qwen2.5-3B-Instruct

Light writing · emails · weekly reports · drafts

AlibabaCN

Size2.0 GB

Min RAM8 GB

VRAM3 GB opt

CPU SpeedFast

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

Qwen2.5-7B-Instruct

Best-value Chinese writing · docs · plans · copy (2 parts)

AlibabaCNPopular

Size4.47 GB

Min RAM16 GB

VRAM6 GB

CPU SpeedMedium

CN Source · HF-MIRROR

⬇ Part 1 ⬇ Part 2 Copy Link

Global Source · HUGGING FACE

⬇ Part 1 ⬇ Part 2 Copy Link

Qwen3-4B

New-gen Qwen · stronger instruction following · efficient

AlibabaCN

Size2.38 GB

Min RAM8 GB

VRAM3 GB opt

CPU SpeedFast

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

Qwen3-8B

New-gen Qwen · all-round · writing & office

AlibabaCN

Size4.79 GB

Min RAM16 GB

VRAM6 GB

CPU SpeedMedium

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

GLM-4-9B-Chat

Long-doc summary · long workflows · complex tasks

Zhipu AICN

Size5.96 GB

Min RAM16 GB

VRAM8 GB

CPU SpeedMedium

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

Llama-3.1-8B-Instruct

Meta open flagship · strong English & code · general

MetaUSA

Size4.69 GB

Min RAM16 GB

VRAM6 GB

CPU SpeedMedium

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

Phi-4-mini-Instruct

Microsoft mini · strong reasoning & math · low-spec

MicrosoftUSA

Size2.38 GB

Min RAM8 GB

VRAM3 GB opt

CPU SpeedFast

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

Mistral-Nemo-12B-Instruct

France Mistral × NVIDIA · 128K context · multilingual (incl. Chinese)

MistralFrance

Size7.62 GB

Min RAM16 GB

VRAM8 GB

CPU SpeedMedium

CN · HF-MIRROR

⬇ Download Copy link

GLOBAL · HUGGING FACE

⬇ Download Copy link

Falcon3-7B-Instruct

UAE TII · multilingual · low-spec friendly

TIIUAE

Size4.66 GB

Min RAM16 GB

VRAM6 GB

CPU SpeedMedium

CN · HF-MIRROR

⬇ Download Copy link

GLOBAL · HUGGING FACE

⬇ Download Copy link

SOLAR-10.7B-Instruct-v1.0

Korea Upstage · balanced all-rounder · Korean/English

UpstageKorea

Size6.28 GB

Min RAM16 GB

VRAM6 GB

CPU SpeedMedium

CN · HF-MIRROR

⬇ Download Copy link

GLOBAL · HUGGING FACE

⬇ Download Copy link

PLaMo-13B-Instruct

Japan Preferred Networks · Japanese / multilingual

PFNJapan

Size8.17 GB

Min RAM16 GB

VRAM8 GB

CPU SpeedSlow

CN · HF-MIRROR

⬇ Download Copy link

GLOBAL · HUGGING FACE

⬇ Download Copy link

Qwen2.5-VL-7B

Qwen vision · image understanding · doc OCR · screenshots

AlibabaCN

Size4.47 GB

Min RAM16 GB

VRAM6 GB

CPU SpeedMedium

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

MiniCPM-V-2.6

OpenBMB vision VLM · image Q&A · low VRAM

OpenBMBCN

Size4.46 GB

Min RAM16 GB

VRAM6 GB

CPU SpeedMedium

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

Gemma-3-4B-it

Google multimodal · image understanding · low-spec

GoogleUSA

Size2.37 GB

Min RAM8 GB

VRAM3 GB opt

CPU SpeedFast

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

Gemma-3-12B-it

Google multimodal · enhanced image understanding

GoogleUSA

Size6.96 GB

Min RAM16 GB

VRAM8 GB

CPU SpeedSlow

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

DeepSeek-R1-Distill-1.5B

Light reasoning · logic assistant · low-spec

DeepSeekCN

Size1.07 GB

Min RAM8 GB

VRAM2 GB opt

CPU SpeedFast

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

DeepSeek-R1-Distill-7B

DeepSeek reasoning distilled · logic analysis · math

DeepSeekCN

Size4.47 GB

Min RAM16 GB

VRAM6 GB

CPU SpeedMedium

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

DeepSeek-R1-Distill-14B

Reasoning distilled 14B · deeper analysis · 32GB tier

DeepSeekCN

Size8.57 GB

Min RAM32 GB

VRAM10 GB

CPU SpeedSlow

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link

QwQ-32B

Qwen deep reasoning · top-tier open reasoning · high-end

AlibabaCN

Size18.9 GB

Min RAM32 GB

VRAM24 GB

CPU SpeedVery Slow

CN Source · HF-MIRROR

⬇ Download Copy Link

Global Source · HUGGING FACE

⬇ Download Copy Link