🚨码住!Qwen / DeepSeek / GLM / MiniMax / Ling 本地部署最强合集!
云端随时能被砍、涨价、加门、改规则,但下到你硬盘上的文件谁也拿不走!
🔵DeepSeek 系列
DeepSeek V4.1 Flash 1017(552B MoE + Engram,1M上下文)
•Unsloth Dynamic V3.1 GGUF(1-bit起):https://huggingface.co/unsloth/DeepSeek-V4.1-Flash-1017-GGUF
🟣Qwen 系列
Qwen3.8-27B(27B稠密,视觉+推理,256K)
•Unsloth Dynamic V3.1 GGUF:https://huggingface.co/unsloth/Qwen3.8-27B-GGUF
•bartowski:https://huggingface.co/bartowski/Qwen3.8-27B-GGUF
•Abliterated:https://huggingface.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF
⚪️Qwen3.8 Flash-Next(MoE,本地可跑)
•Unsloth Dynamic:https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF
•bartowski:https://huggingface.co/bartowski/Qwen3.8-Flash-Next-GGUF
⚫️GLM 系列
GLM-5.3-Flash(320B / 18B激活,原生多模态)
•Unsloth Dynamic:https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF
•6block:https://huggingface.co/6block/GLM-5.3-Flash-GGUF
•Abliterated:https://huggingface.co/huihui-ai/Huihui-GLM-5.3-Flash-abliterated-GGUF
🟡MiniMax 系列
MiniMax-M3.1-Flash(428B / 23B激活,1M上下文,多模态)
•官方权重:https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash
•Unsloth Dynamic GGUF(1-bit~Q8,推荐):https://huggingface.co/unsloth/MiniMax-M3.1-Flash-GGUF
•smalinin Q2_K:https://huggingface.co/smalinin/MiniMax-M3.1-Flash-Q2_K-GGUF
🟢Ling 系列
Ling-3.1-Flash(560B / 25B激活,混合MoE,MIT)
•官方权重:https://huggingface.co/inclusionAI/Ling-3.1-flash
•bartowski 1019量化(IQ1到Q8):https://huggingface.co/bartowski/Ling-3.1-flash-1019-GGUF
•Abliterated:https://huggingface.co/mrfakename/Ling-3.1-flash-1019-abliterated-GGUF
下载提示:优先选 Unsloth / bartowski 的 Dynamic 或标准量化,用
hf download OWNER/REPO --include "*Q4_K_M*" --local-dir ./MODEL
大模型注意磁盘和内存,Engram/专家可SSD卸载
现在就落盘,文件在你硬盘上,规则再变也拿不走。
#LocalAI #Qwen #DeepSeek #GLM #MiniMax #Ling #开源模型 #本地部署 #GGUF #Unsloth #Abliterated