What are the hardware requirements for UI UG2 4B?

UI UG2 4B requires approximately 8.9 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed UI UG2 4B and how large is it?

UI UG2 4B is developed by afx-team, a model with 4b 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 UI UG2 4B?

UI UG2 4B is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

UI UG2 4B by afx-team — VRAM 8.9GB

Name: UI UG2 4B
Author: afx-team

UI UG2 4B is an open-source language model by afx-team. Features: 4b LLM, VRAM: 8.9GB, License: apache-2.0, LLM Explorer Score: 0.26.

Arxiv:2509.24361 Conversational Endpoints compatible Image-text-to-text Mllm Qwen3 vl Region:us Safetensors Sharded Tensorflow Ui Ui-ug Ui-ug2

Model Card on HF 🤗: https://huggingface.co/afx-team/UI-UG2-4B

UI UG2 4B Parameters and Internals

LLM Name	UI UG2 4B
Repository 🤗	https://huggingface.co/afx-team/UI-UG2-4B
Model Size	4b
Required VRAM	8.9 GB
Updated	2026-11-06
Maintainer	afx-team
Model Type	qwen3_vl
Model Files	5.0 GB: 1-of-2 3.9 GB: 2-of-2
Model Architecture	Qwen3VLForConditionalGeneration
License	apache-2.0
Model Max Length	262144
Transformers Version	4.57.0
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Errors	replace

Rank the UI UG2 4B 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 UI UG2 4B

Best Alternatives	Context / RAM	Downloads	Likes
WeVisOne VL 4B	0K / 19.3 GB	2605	58
Qwen3 VL 4B Instruct	0K / 8.9 GB	3611570	437
MinerU Popo	0K / 17.6 GB	26895	4
Qwen3 VL 4B Instruct FP8	0K / 6.1 GB	301103	65
WeVisDoc 4B	0K / 9.7 GB	349	31
Qwen3 VL 4B Thinking Pashto Zi	0K / 8.9 GB	1426	0
VIG Qwen3 VL Thinking 4B	0K / 9.7 GB	48	1
AA VL 4B	0K / 8.9 GB	87	0
Qwen3 VL 4B VQA SFT	0K / 8.9 GB	40	0
Qwen3 VL 4B Thinking	0K / 8.9 GB	21715	118

Note: green Score (e.g. "73.2") means that the model is better than afx-team/UI-UG2-4B.