This model is a fine-tuned version of Qwen/Qwen3.5-4B-Base on the Omni-Edu-70K dataset. The following hyperparameters were used during training: - learningrate: 5e-06 - trainbatchsize: 1 - evalbatchsize: 8 - distributedtype: multi-GPU - numdevices: 8 - gradientaccumulationsteps: 8 - totaltrainbatchsize: 64 - totalevalbatchsize: 64 - lrschedulertype: cosine - lrschedulerwarmupsteps: 0.1 - numepochs: 3.0 - Transformers 5.2.0 - Pytorch 2.10.0 - Datasets 4.0.0 - Tokenizers 0.22.2
Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.
Parameters4.4B
Context262,144
Weights8.9 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads3.8M
Runs On
What it takes to serve Qwen3-VL-4B-Instruct (4.4B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 8.9 GB | 10.7 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 4.4 GB | 5.3 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 2.2 GB | 2.7 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
SAVRN's Notes on Qwen3-VL-4B-Instruct
Feed it an image and a question and it answers in text. At 4.4B parameters the memory figures put it on shared hardware: 2.7 GB needed at 4-bit, 5.3 GB at 8-bit, 10.7 GB for the full 16-bit weights. The cheapest Index setup is one MI300X with 192 GB at $1.85 an hour on-demand, a price set by the card rather than the model, so a single copy leaves most of that card idle.
Apache 2.0 permits commercial use, modification and redistribution, so a fine-tuned version can ship inside a product if the license and copyright notices travel with it and significant changes are stated. Confirm the 262,144-token context length, the ceiling on how much image and text goes into one request, and check that the files you receive match the October 11, 2025 release. Four linked technical reports carry the lineage back to the Qwen-VL paper.
Model Card
By Qwen, published under apache-2.0, revision ebb281ec70b0.
Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.
This generation delivers comprehensive upgrades across the board: superior text understanding & generation, deeper visual perception & reasoning, extended context length, enhanced spatial and video dynamics comprehension, and stronger agent interaction capabilities.
Available in Dense and MoE architectures that scale from edge to cloud, with Instruct and reasoning‑enhanced Thinking editions for flexible, on‑demand deployment.
Key Enhancements:
Configuration
- Architecture
- Qwen3VLForConditionalGeneration
- Context length (tokens)
- 262,144
- Layers
- 36
- Hidden size
- 2,560
- Feed-forward size
- 9,728
- Attention heads
- 32
- Key/value heads
- 8
- Head dimension
- 128
- Vocabulary size
- 151,936
- RoPE base
- 5,000,000
- Model type
- qwen3_vl
Identity and Version
- Repository
- Qwen/Qwen3-VL-4B-Instruct
- Publisher
- Qwen
- Task
- Image and text to text
- Modality
- Image and text
- Library
- transformers
- Parameters
- 4.4B parameters
- Languages
- Not stated by the source
- Revision
- ebb281ec70b05090aa6165b016eac8ec08e71b17
- First published
- 2025-10-11
- Last updated
- 2025-10-15
Files and Weights
14 files, 8.9 GB in total. The weights are 2 files totalling 8.9 GB in safetensors.
Weights2 files · 8.9 GB
Configuration6 files · 72.8 KB
Tokenizer4 files · 11.5 MB
Documentation1 file · 7.1 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model-00001-of-00002.safetensors | Weights | 5.0 GB | 30a01a055662 |
| model-00002-of-00002.safetensors | Weights | 3.9 GB | 046296a2a387 |
| chat_template.json | Configuration | 5.5 KB | — |
| config.json | Configuration | 1.5 KB | — |
| generation_config.json | Configuration | 269 B | — |
| model.safetensors.index.json | Configuration | 64.7 KB | — |
| preprocessor_config.json | Configuration | 390 B | — |
| video_preprocessor_config.json | Configuration | 385 B | — |
| README.md | Documentation | 7.1 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
| merges.txt | Tokenizer | 1.7 MB | — |
| tokenizer.json | Tokenizer | 7.0 MB | — |
| tokenizer_config.json | Tokenizer | 10.9 KB | — |
| vocab.json | Tokenizer | 2.8 MB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 8.9 GB
Download from Qwen
Released by Qwen through ModelScope. Read the license.
Built From
- Described by arXiv:2308.12966
- Described by arXiv:2409.12191
- Described by arXiv:2502.13923
- Described by arXiv:2505.09388
Evaluations
Each result is shown as reported, with the conditions its reporter stated. None is a SAVRN measurement. A comparison lines two results up only when their configuration, unit and setup are all stated and identical.
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| tiiuae/PBench | Task averageMetric averageSetup Detection-only model combined with SAM2 to convert boxes to segmentation masks.Comparison conditions not established | 49.2 | Community Evals Reported by a third party |
Evaluated revision not stated | 2026-05-11 |
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 8.9 GB |
| 16-bit | 8.9 GB |
| 8-bit | 4.4 GB |
| 4-bit | 2.2 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Compare Qwen3-VL-4B-Instruct
Questions About Qwen3-VL-4B-Instruct
How much GPU memory does Qwen3-VL-4B-Instruct need?
About 10.7 GB at 16-bit and 2.7 GB at 4-bit: the weights (4.4B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Qwen3-VL-4B-Instruct on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use Qwen3-VL-4B-Instruct commercially?
Yes. Qwen3-VL-4B-Instruct is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
What is Qwen3-VL-4B-Instruct's context length?
262,144 tokens, from the maximum position embeddings in its published configuration.
Similar Models
[Gemma 3 Technical Report][g3-tech-report] [Responsible Generative AI Toolkit][rai-toolkit] [Gemma on Kaggle][kaggle-gemma] [Gemma on Vertex Model Garden][vertex-mg-gemma3] Summary description and brief definition of inputs and outputs. Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. Gemma 3 models are multimodal, handling text and image input and generating text output, with open weights for both pre-trained variants and instruction-tuned variants. Gemma 3 has a large, 128K context window, multilingual support in over 140 languages, and is available in more sizes than previous…
Model on Google Cloud Model Garden: MedGemma GitHub repository (supporting code, Colab notebooks, discussions, and Foundations terms of use](https://developers.google.com/health-ai-developer-foundations/terms). This section describes the MedGemma model and how to use it. MedGemma is a collection of Gemma 3 variants that are trained for performance on medical text and image comprehension. Developers can use MedGemma to accelerate building healthcare-based AI applications. MedGemma currently comes in three variants: a 4B multimodal version and 27B text-only and multimodal versions. Both MedGemma multimodal versions utilize a SigLIP image encoder that has been specifically pre-trained on a…
Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after difficulty labeling and rebalancing; we report the averaged scores on 55 languages using XCOMET-XXL. Empty cells (--) indicate scores not yet available or not applicable. Empty cells (--) indicate scores not…
This model is Rex-Omni, a 3B-parameter Multimodal Large Language Model (MLLM) presented in the paper "Detect Anything via Next Point Prediction". It is compatible with the Hugging Face transformers library and is licensed under the IDEA License 1.0. src="https://img.shields.io/badge/RexOmni-Website-BADFDB?style=flat-square&logo=deno&logoColor=violet&color=BADFDB" alt="RexThinker Website" src="https://img.shields.io/badge/RexOmni-Paper-Red%25red?logo=arxiv&logoColor=red&color=yellow" alt="RexThinker Paper on arXiv" src="https://img.shields.io/badge/RexOmni-Weight-orange?logo=huggingface&logoColor=yellow" alt="RexThinker weight on Hugging Face"…
BLIP-2 model, leveraging Flan T5-xl (a large language model). It was introduced in the paper BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models by Li et al. and first released in this repository. Disclaimer: The team releasing BLIP-2 did not write a model card for this model so this model card has been written by the Hugging Face team. BLIP-2 consists of 3 models: a CLIP-like image encoder, a Querying Transformer (Q-Former) and a large language model. The authors initialize the weights of the image encoder and large language model from pre-trained checkpoints and keep them frozen while training the Querying Transformer, which is a…