What are the hardware requirements for Qwen3 Next 80B A3B Thinking?

Qwen3 Next 80B A3B Thinking requires approximately 162.7 GB of VRAM and supports a context window of 256K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Qwen3 Next 80B A3B Thinking and how large is it?

Qwen3 Next 80B A3B Thinking is developed by Qwen, a model with 80b 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 Next 80B A3B Thinking?

Qwen3 Next 80B A3B Thinking is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Qwen3 Next 80B A3B Thinking by Qwen — VRAM 162.7GB, 256K context

Name: Qwen3 Next 80B A3B Thinking
Author: Qwen

Qwen3 Next 80B A3B Thinking is an open-source language model by Qwen. Features: 80b LLM, VRAM: 162.7GB, Context: 256K, License: apache-2.0, LLM Explorer Score: 0.29.

Arxiv:2309.00071 Arxiv:2501.15383 Arxiv:2505.09388 Conversational Deploy:azure Endpoints compatible Eval-results Qwen3 next Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Thinking

Qwen3 Next 80B A3B Thinking Parameters and Internals

LLM Name	Qwen3 Next 80B A3B Thinking
Repository 🤗	https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Thinking
Model Size	80b
Required VRAM	162.7 GB
Updated	2026-07-28
Maintainer	Qwen
Model Type	qwen3_next
Model Files	4.0 GB: 1-of-41 4.0 GB: 2-of-41 4.0 GB: 3-of-41 4.0 GB: 4-of-41 4.0 GB: 5-of-41 4.0 GB: 6-of-41 4.0 GB: 7-of-41 4.0 GB: 8-of-41 4.0 GB: 9-of-41 4.0 GB: 10-of-41 4.0 GB: 11-of-41 4.0 GB: 12-of-41 4.0 GB: 13-of-41 4.0 GB: 14-of-41 4.0 GB: 15-of-41 4.0 GB: 16-of-41 4.0 GB: 17-of-41 4.0 GB: 18-of-41 4.0 GB: 19-of-41 4.0 GB: 20-of-41 4.0 GB: 21-of-41 4.0 GB: 22-of-41 4.0 GB: 23-of-41 4.0 GB: 24-of-41 4.0 GB: 25-of-41 4.0 GB: 26-of-41 4.0 GB: 27-of-41 4.0 GB: 28-of-41 4.0 GB: 29-of-41 4.0 GB: 30-of-41 4.0 GB: 31-of-41 4.0 GB: 32-of-41 4.0 GB: 33-of-41 4.0 GB: 34-of-41 4.0 GB: 35-of-41 4.0 GB: 36-of-41 4.0 GB: 37-of-41 4.0 GB: 38-of-41 4.0 GB: 39-of-41 3.4 GB: 40-of-41 3.3 GB: 41-of-41
Model Architecture	Qwen3NextForCausalLM
License	apache-2.0
Context Length	262144
Model Max Length	262144
Transformers Version	4.57.0.dev0
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	151936
Torch Data Type	bfloat16
Errors	replace

Rank the Qwen3 Next 80B A3B Thinking 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

Quantized Models of the Qwen3 Next 80B A3B Thinking

Model	Likes	Downloads	VRAM
...Next 80B A3B Thinking AWQ 4bit	23	106058	49 GB
...Next 80B A3B Thinking AWQ 4bit	11	36149	47 GB
...en3 Next 80B A3B Thinking 4bit	4	3953	44 GB
...en3 Next 80B A3B Thinking 8bit	3	2587	83 GB

Best Alternatives to Qwen3 Next 80B A3B Thinking

Best Alternatives	Context / RAM	Downloads	Likes
Qwen3 Next 80B A3B Instruct	256K / 162.7 GB	422	89
Qwen3 Next 80B A3B Thinking	256K / 162.7 GB	100	7
Qwen3 Next 80B A3B Instruct	256K / 162.7 GB	286679	1042
...wen3 Next 80B A3B Instruct FP8	256K / 81.8 GB	329583	89
Qwen3 Next MoE	256K / 0 GB	127090	4
...n3 Next 80B A3B Instruct NVFP4	256K / 50.7 GB	13898	39
...wen3 Next 80B A3B Thinking FP8	256K / 81.8 GB	20135	54
...n3 Next 80B A3B Thinking NVFP4	256K / 50.7 GB	2278	64
...t 80B A3B Instruct FP8 Dynamic	256K / 80.5 GB	506	4
...ext 80B A3B Instruct Mxfp4 Mlx	256K / 42 GB	243	8

Note: green Score (e.g. "73.2") means that the model is better than Qwen/Qwen3-Next-80B-A3B-Thinking.