Qwen3-VL-4B-Instruct

Name: Qwen3-VL-4B-Instruct
Author: Alibaba

Multimodal modelQwen3

Release date: 2025-10-15Updated: 2025-10-15 08:23:571,302

Live demoGitHub Hugging Face Compare

Parameters

Context length

256K

Chinese support

Not supported

Reasoning ability

Qwen3-VL-4B-Instruct is an AI model published by Alibaba, released on 2025-10-15, for Multimodal model, with 4B parameters, and 256K context length, requiring about 8.89 GB storage, with a 95.30 score on DocVQA.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Qwen3-VL-4B-Instruct

Model basics

Reasoning traces

Not supported

Thinking modes

Thinking modes not supported

Context length

256K tokens

Max output length

No data

Model type

Multimodal model

Modality (in / out)

Text, Image, Video → Text

Release date

2025-10-15

Model file size

8.89 GB

MoE architecture

Total params / Active params

4B / N/A

Knowledge cutoff

No data

Qwen3-VL-4B-Instruct

Open source & experience

Code license

Apache 2.0

Weights license

Apache 2.0- Commercial use permitted

GitHub repo

https://github.com/QwenLM/Qwen3-VL

Hugging Face

https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct

Live demo

No live demo

Qwen3-VL-4B-Instruct

Official resources

Paper

DataLearnerAI blog

No blog post yet

Qwen3-VL-4B-Instruct

API details

API speed

3/5

No public API pricing yet.

Qwen3-VL-4B-Instruct

Benchmark Results

Qwen3-VL-4B-Instruct currently shows benchmark results led by DocVQA (3 / 5, score 95.30), MMMU (26 / 29, score 67.40). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.