What are the hardware requirements for QVQ 72B Preview Bf16?

QVQ 72B Preview Bf16 requires approximately 147.7 GB of VRAM and supports a context window of 125K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed QVQ 72B Preview Bf16 and how large is it?

QVQ 72B Preview Bf16 is developed by mlx-community, a model with 72b 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 QVQ 72B Preview Bf16?

QVQ 72B Preview Bf16 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

QVQ 72B Preview Bf16 by mlx-community — VRAM 147.7GB, 125K context

Name: QVQ 72B Preview Bf16
Author: mlx-community

QVQ 72B Preview Bf16 is an open-source language model by mlx-community. Features: 72b LLM, VRAM: 147.7GB, Context: 125K, License: other, LLM Explorer Score: 0.16.

Base model:finetune:qwen/qwen2... Base model:qwen/qwen2-vl-72b Chat Conversational En Endpoints compatible Image-text-to-text Mlx Qwen2 vl Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/mlx-community/QVQ-72B-Preview-bf16

QVQ 72B Preview Bf16 Parameters and Internals

LLM Name	QVQ 72B Preview Bf16
Repository 🤗	https://huggingface.co/mlx-community/QVQ-72B-Preview-bf16
Base Model(s)	Qwen2 VL 72B Qwen/Qwen2-VL-72B
Model Size	72b
Required VRAM	147.7 GB
Updated	2026-08-08
Maintainer	mlx-community
Model Type	qwen2_vl
Model Files	5.2 GB: 1-of-28 5.3 GB: 2-of-28 5.3 GB: 3-of-28 5.3 GB: 4-of-28 5.3 GB: 5-of-28 5.3 GB: 6-of-28 5.3 GB: 7-of-28 5.3 GB: 8-of-28 5.3 GB: 9-of-28 5.3 GB: 10-of-28 5.3 GB: 11-of-28 5.3 GB: 12-of-28 5.3 GB: 13-of-28 5.3 GB: 14-of-28 5.3 GB: 15-of-28 5.3 GB: 16-of-28 5.3 GB: 17-of-28 5.3 GB: 18-of-28 5.3 GB: 19-of-28 5.3 GB: 20-of-28 5.3 GB: 21-of-28 5.3 GB: 22-of-28 5.3 GB: 23-of-28 5.3 GB: 24-of-28 5.3 GB: 25-of-28 5.3 GB: 26-of-28 5.3 GB: 27-of-28 4.7 GB: 28-of-28
Supported Languages	en
Model Architecture	Qwen2VLForConditionalGeneration
License	other
Context Length	128000
Model Max Length	128000
Transformers Version	4.41.2
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	152064
Torch Data Type	bfloat16
Errors	replace

Rank the QVQ 72B Preview Bf16 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 QVQ 72B Preview Bf16

Best Alternatives	Context / RAM	Downloads	Likes
QVQ 72B Preview	125K / 146.8 GB	8572	610
Qwen2 VL 72B Instruct	32K / 146.8 GB	60157	311
UI TARS 72B DPO	32K / 147.5 GB	522	159
UI TARS 72B DPO Bf16	32K / 147.7 GB	30	1
UI TARS 72B SFT Bf16	32K / 147.7 GB	32	1
UI TARS 72B SFT	32K / 204.2 GB	110	25
Qwen2 VL 72B	32K / 146.8 GB	151	80
Qwen2 VL 72B Instruct	32K / 147.5 GB	61	1
QVQ 72B Preview 4bit	125K / 41.3 GB	20	7
QVQ 72B Preview 6bit	125K / 59.7 GB	34	2

Note: green Score (e.g. "73.2") means that the model is better than mlx-community/QVQ-72B-Preview-bf16.