What are the hardware requirements for Qwen3.8 Flash Next MLX 4bit MTP?

Qwen3.8 Flash Next MLX 4bit MTP requires approximately 112.8 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Qwen3.8 Flash Next MLX 4bit MTP and how large is it?

Qwen3.8 Flash Next MLX 4bit MTP is developed by TensorFold, a model with 180b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Qwen3.8 Flash Next MLX 4bit MTP?

Qwen3.8 Flash Next MLX 4bit MTP is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Qwen3.8 Flash Next MLX 4bit MTP by TensorFold — VRAM 112.8GB

Name: Qwen3.8 Flash Next MLX 4bit MTP
Author: TensorFold

Qwen3.8 Flash Next MLX 4bit MTP is an open-source language model by TensorFold. Features: 180b LLM, VRAM: 112.8GB, License: other, Quantized.

4-bit 4bit Apple-silicon Base model:quantized:qwen/qwen... Base model:qwen/qwen3.8-flash-... Conversational Image-text-to-text Mixture-of-experts Mlx Mlx-vlm Mtp Omlx Quantized Qwen Qwen3.8 Qwen4-exp Qwen4 exp Region:us Safetensors Sharded Speculative-decoding Tensorflow Vision Vision-language

Model Card on HF 🤗: https://huggingface.co/TensorFold/Qwen3.8-Flash-Next-MLX-4bit-MTP

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

LLM Name	Qwen3.8 Flash Next MLX 4bit MTP
Repository 🤗	https://huggingface.co/TensorFold/Qwen3.8-Flash-Next-MLX-4bit-MTP
Base Model(s)	Qwen/Qwen3.8-Flash-Next Qwen/Qwen3.8-Flash-Next
Model Size	180b
Required VRAM	112.8 GB
Updated	2026-10-05
Maintainer	TensorFold
Model Type	qwen4_exp
Model Files	5.3 GB: 1-of-22 5.2 GB: 2-of-22 5.2 GB: 3-of-22 5.2 GB: 4-of-22 5.2 GB: 5-of-22 5.2 GB: 6-of-22 5.1 GB: 7-of-22 5.3 GB: 8-of-22 5.3 GB: 9-of-22 5.0 GB: 10-of-22 5.3 GB: 11-of-22 5.0 GB: 12-of-22 5.3 GB: 13-of-22 5.3 GB: 14-of-22 5.0 GB: 15-of-22 5.3 GB: 16-of-22 5.0 GB: 17-of-22 5.3 GB: 18-of-22 5.3 GB: 19-of-22 5.0 GB: 20-of-22 5.3 GB: 21-of-22 3.7 GB: 22-of-22
Quantization Type	4bit
Model Architecture	Qwen4ExpForConditionalGeneration
License	other
Model Max Length	262144
Transformers Version	5.8.0.dev0
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Errors	replace

Rank the Qwen3.8 Flash Next MLX 4bit MTP capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Qwen3.8 Flash Next MLX 4bit MTP

Best Alternatives	Context / RAM	Downloads	Likes
...3.8 Flash Next OQ3.5e Fp16 Mtp	0K / 90.7 GB	766	1
Qwen3.8 Flash Next 4bit Mtp	0K / 112.9 GB	540	3
...en3.8 Flash Next OQ4e Fp16 Mtp	0K / 107.3 GB	746	2
...wen3.8 Flash Next MLX 4bit MTP	0K / 112.8 GB	1643	5
...MLX Activation 3bit Native PLE	0K / 34.7 GB	284	1
Qwen3.8 Flash Next	0K / 129.5 GB	4810	3918
Darwin 180B RSI	0K / 119 GB	353	94
Qwen3.8 Flash Next FP8	0K / 65.7 GB	2219	117
Tinfield 1	0K / 129.5 GB	577	41
CYBER FROST 3.8 BF16	0K / 124.3 GB	1047	42

Note: green Score (e.g. "73.2") means that the model is better than TensorFold/Qwen3.8-Flash-Next-MLX-4bit-MTP.