LLM EXPLORER 65,695 MODELS INDEXED

Qwen3.8 Flash Next GSQ RCO Coder MLX by mbrown-uno

By mbrown-uno · 490 downloads
HF Repo ↗ Copy URL

Qwen3.8 Flash Next GSQ RCO Coder MLX is an open-source language model by mbrown-uno. Features: 119.6b LLM, VRAM: 81.5GB, License: apache-2.0, Code Generating.

  Arxiv:2604.18556   Arxiv:2605.00649   4-bit   Agentic   Apple-silicon Base model:ista-daslab/qwen3.8... Base model:quantized:ista-dasl...   Code   Codegen   Conversational   En   Expert-pruning   Image-text-to-text   Mlx   Moe   Mtp   Omlx   Qwen3.8   Qwen4 exp   Region:us   Safetensors   Sharded   Speculative-decoding   Tensorflow   Tool-calling
Model Card on HF 🤗: https://huggingface.co/mbrown-uno/Qwen3.8-Flash-Next-GSQ-RCO-Coder-MLX 

Qwen3.8 Flash Next GSQ RCO Coder MLX Parameters and Internals

LLM NameQwen3.8 Flash Next GSQ RCO Coder MLX
Repository 🤗https://huggingface.co/mbrown-uno/Qwen3.8-Flash-Next-GSQ-RCO-Coder-MLX 
Base Model(s)  ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-Coder-GGUF   Qwen/Qwen3.8-Flash-Next   ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-Coder-GGUF   Qwen/Qwen3.8-Flash-Next
Model Size119.6b
Required VRAM81.5 GB
Updated2026-10-27
Maintainermbrown-uno
Model Typeqwen4_exp
Model Files  5.5 GB: 1-of-16   5.4 GB: 2-of-16   5.4 GB: 3-of-16   5.5 GB: 4-of-16   5.4 GB: 5-of-16   5.5 GB: 6-of-16   5.6 GB: 7-of-16   5.4 GB: 8-of-16   5.4 GB: 9-of-16   5.4 GB: 10-of-16   5.4 GB: 11-of-16   5.4 GB: 12-of-16   5.4 GB: 13-of-16   5.4 GB: 14-of-16   4.7 GB: 15-of-16   0.7 GB: 16-of-16
Supported Languagesen
Generates CodeYes
Model ArchitectureQwen4ExpForConditionalGeneration
Licenseapache-2.0
Model Max Length262144
Transformers Version5.8.0.dev0
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Errorsreplace