What are the hardware requirements for Deepseek Vl2?

Deepseek Vl2 requires approximately 55 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Deepseek Vl2 and how large is it?

Deepseek Vl2 is developed by deepseek-ai, a model with 27.5b 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 Deepseek Vl2?

Deepseek Vl2 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Deepseek Vl2 by deepseek-ai — VRAM 55GB

Name: Deepseek Vl2
Author: deepseek-ai

Deepseek Vl2 is an open-source language model by deepseek-ai. Features: 27.5b LLM, VRAM: 55GB, License: other.

Arxiv:2412.10302 Deepseek vl v2 Endpoints compatible Image-text-to-text Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/deepseek-ai/deepseek-vl2

Deepseek Vl2 Parameters and Internals

LLM Name	Deepseek Vl2
Repository 🤗	https://huggingface.co/deepseek-ai/deepseek-vl2
Model Size	27.5b
Required VRAM	55 GB
Updated	2026-08-10
Maintainer	deepseek-ai
Model Type	deepseek_vl_v2
Model Files	8.6 GB: 1-of-8 8.6 GB: 2-of-8 8.6 GB: 3-of-8 1.4 GB: 4-of-8 8.6 GB: 5-of-8 8.6 GB: 6-of-8 8.6 GB: 7-of-8 2.0 GB: 8-of-8
Model Architecture	AutoModel
License	other
Transformers Version	4.38.2
Tokenizer Class	LlamaTokenizer
Padding Token	<｜▁pad▁｜>
Torch Data Type	bfloat16

Rank the Deepseek Vl2 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