NewHorizonGroup/Qwen3.8-27B-oQ4-mtp

library_name: mlx tags:

• —mlx
• —oq
• —quantized license: mit language:
• —zh
• —en
• —ja
• —ko base_model:
• —Qwen/Qwen3.8-27B pipelinetag: image-text-to-text --- 

Qwen3.8-27B-oQ4-mtp

This model was quantized using oQ (oMLX v0.5.7) mixed-precision quantization.

jundot is a developer I deeply admire. He said:

oQ: oMLX Universal Dynamic Quantization

Quantization should not be exclusive to any particular inference server. oQ produces standard mlx-lm models that work everywhere — oMLX, mlx-lm, LM Studio, and any app that supports the MLX safetensors format. No custom loader required.

A tribute to him.🍺

Quantization details

• —Model type: qwen3_5
• —Bits: 4
• —Group size: 64
• —Format: MLX safetensors