LLM EXPLORER 60,062 MODELS INDEXED

Qwen3.8 Flash Next Abliterated MLX 4bit by grant-ai

By grant-ai · 866 downloads
HF Repo ↗ Copy URL

Qwen3.8 Flash Next Abliterated MLX 4bit is an open-source language model by grant-ai. Features: 34.3b LLM, VRAM: 113.4GB, License: other, Quantized, LLM Explorer Score: 0.33.

  4-bit   4bit   Abliterated   Apple-silicon Base model:blackfrost-ai/qwen3... Base model:quantized:blackfros...   Conversational   Derisked   Image-text-to-text   Mlx   Mtp   Omlx   Quantized   Qwen3.8-flash-next   Qwen4 exp   Region:us   Safetensors   Sharded   Tensorflow   Uncensored
Model Card on HF 🤗: https://huggingface.co/grant-ai/Qwen3.8-Flash-Next-Abliterated-MLX-4bit 

Qwen3.8 Flash Next Abliterated MLX 4bit Parameters and Internals

LLM NameQwen3.8 Flash Next Abliterated MLX 4bit
Repository 🤗https://huggingface.co/grant-ai/Qwen3.8-Flash-Next-Abliterated-MLX-4bit 
Base Model(s)  Qwen/Qwen3.8-Flash-Next   Blackfrost-AI/Qwen3.8-Flash-Next-DERISKED-BF16   Qwen/Qwen3.8-Flash-Next   Blackfrost-AI/Qwen3.8-Flash-Next-DERISKED-BF16
Model Size34.3b
Required VRAM113.4 GB
Updated2026-08-31
Maintainergrant-ai
Model Typeqwen4_exp
Model Files  5.0 GB: 1-of-23   5.0 GB: 2-of-23   5.0 GB: 3-of-23   5.0 GB: 4-of-23   5.0 GB: 5-of-23   5.0 GB: 6-of-23   5.0 GB: 7-of-23   5.1 GB: 8-of-23   5.2 GB: 9-of-23   5.2 GB: 10-of-23   5.2 GB: 11-of-23   5.2 GB: 12-of-23   5.2 GB: 13-of-23   5.2 GB: 14-of-23   5.2 GB: 15-of-23   5.2 GB: 16-of-23   5.2 GB: 17-of-23   5.2 GB: 18-of-23   5.2 GB: 19-of-23   5.2 GB: 20-of-23   5.2 GB: 21-of-23   5.2 GB: 22-of-23   0.5 GB: 23-of-23
Quantization Type4bit
Model ArchitectureQwen4ExpForConditionalGeneration
Licenseother
Model Max Length262144
Transformers Version5.8.0.dev0
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Errorsreplace