What are the hardware requirements for Vintern 1B V2?

Vintern 1B V2 requires approximately 1.9 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Vintern 1B V2 and how large is it?

Vintern 1B V2 is developed by 5CD-AI, a model with 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 Vintern 1B V2?

Vintern 1B V2 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Vintern 1B V2 by 5CD-AI — VRAM 1.9GB

Name: Vintern 1B V2
Author: 5CD-AI

Vintern 1B V2 is an open-source language model by 5CD-AI. Features: 1b LLM, VRAM: 1.9GB, License: mit, LLM Explorer Score: 0.15.

Arxiv:2404.07922 Arxiv:2404.16821 Arxiv:2407.10671 Arxiv:2408.12480 Base model:finetune:opengvlab/... Base model:opengvlab/internvl2... Conversational Custom code Dataset:5cd-ai/viet-computersc... Dataset:5cd-ai/viet-doc-vqa Dataset:5cd-ai/viet-doc-vqa-ii Dataset:5cd-ai/viet-geometry-v... Dataset:5cd-ai/viet-handwritin... Dataset:5cd-ai/viet-menu-gemin... Dataset:5cd-ai/viet-ocr-vqa Dataset:5cd-ai/viet-openvivqa-... Dataset:5cd-ai/viet-receipt-vq... Dataset:5cd-ai/viet-resume-vqa Dataset:5cd-ai/viet-sketches-v... Dataset:5cd-ai/viet-vintext-ge... Dataset:5cd-ai/viet-vitextvqa-... Dataset:5cd-ai/viet-wiki-handw... Dataset:vi-vlm/vista En Feature-extraction Image-text-to-text Internvl chat Region:us Safetensors Vi Vision

Model Card on HF 🤗: https://huggingface.co/5CD-AI/Vintern-1B-v2

Vintern 1B V2 Parameters and Internals

LLM Name	Vintern 1B V2
Repository 🤗	https://huggingface.co/5CD-AI/Vintern-1B-v2
Base Model(s)	InternVL2 1B OpenGVLab/InternVL2-1B
Model Size	1b
Required VRAM	1.9 GB
Updated	2026-08-26
Maintainer	5CD-AI
Model Type	internvl_chat
Model Files	1.9 GB
Supported Languages	vi en
Model Architecture	InternVLChatModel
License	mit
Model Max Length	4096
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Torch Data Type	bfloat16
Errors	replace

Rank the Vintern 1B V2 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 Vintern 1B V2

Best Alternatives	Context / RAM	Downloads	Likes
HPD Parsing	0K / 2.1 GB	8362	88
InternVL3 5 1B	0K / 2.1 GB	118776	29
Trendyol Vision Flash	0K / 2.1 GB	78	8
InternVL3 1B	0K / 1.9 GB	420183	85
InternVL2 1B	0K / 1.9 GB	824111	82
InternVL3 5 1B Instruct	0K / 2.1 GB	1133	7
InternVL2 5 1B	0K / 1.9 GB	20818	69
InternVL3 5 1B Flash	0K / 2.3 GB	85	11
InternVL3 5 1B MPO	0K / 2.1 GB	43	2
InternVL3 5 1B Pretrained	0K / 2.1 GB	40	1

Note: green Score (e.g. "73.2") means that the model is better than 5CD-AI/Vintern-1B-v2.