What are the hardware requirements for VLM2Vec Full?

VLM2Vec Full requires approximately 8.3 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed VLM2Vec Full and how large is it?

VLM2Vec Full is developed by TIGER-Lab, a model with 4.1b 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 VLM2Vec Full?

VLM2Vec Full is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

VLM2Vec Full by TIGER-Lab — VRAM 8.3GB, 128K context

Name: VLM2Vec Full
Author: TIGER-Lab

VLM2Vec Full is an open-source language model by TIGER-Lab. Features: 4.1b LLM, VRAM: 8.3GB, Context: 128K, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.24.

Arxiv:2410.05160 Base model:finetune:microsoft/... Base model:microsoft/phi-3.5-v... Conversational Custom code Dataset:tiger-lab/mmeb-train Embedding En Instruct Phi3 v Pytorch Region:us Safetensors Sharded Tensorflow Vision

Model Card on HF 🤗: https://huggingface.co/TIGER-Lab/VLM2Vec-Full

VLM2Vec Full Parameters and Internals

Model Type

multimodal embedding

Additional Notes

The model uses an [EOS] token as the representation of the multimodal inputs.

Supported Languages

en (English)

Training Details

Data Sources:

TIGER-Lab/MMEB-train

Methodology:

Converting an existing well-trained VLM into an embedding model with contrastive learning.

LLM Name	VLM2Vec Full
Repository 🤗	https://huggingface.co/TIGER-Lab/VLM2Vec-Full
Base Model(s)	microsoft/Phi-3.5-vision-instruct microsoft/Phi-3.5-vision-instruct
Model Size	4.1b
Required VRAM	8.3 GB
Updated	2026-08-03
Maintainer	TIGER-Lab
Model Type	phi3_v
Instruction-Based	Yes
Model Files	4.9 GB: 1-of-2 3.4 GB: 2-of-2 4.9 GB: 1-of-2 3.4 GB: 2-of-2
Supported Languages	en
Model Architecture	Phi3VForCausalLM
License	apache-2.0
Context Length	131072
Model Max Length	131072
Transformers Version	4.46.1
Tokenizer Class	LlamaTokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	32064
Torch Data Type	bfloat16

Rank the VLM2Vec Full 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 VLM2Vec Full

Best Alternatives	Context / RAM	Downloads	Likes
Phi 3.5 Vision Instruct	128K / 8.3 GB	1243386	737
Phi 3 Vision 128K Instruct	128K / 8.3 GB	240595	973
Phi 3.5 Vision Instruct Bf16	128K / 8.3 GB	48	2
...on 128K Instruct No Flash Attn	128K / 8.3 GB	13	0
Phi 3 Vision 128K Instruct	128K / 8.3 GB	8	0
...hi 3 HornyVision 128K Instruct	128K / 8.3 GB	17	29
Phi 3 Vision Win Snap	128K / 8.3 GB	12	1

Note: green Score (e.g. "73.2") means that the model is better than TIGER-Lab/VLM2Vec-Full.