Qwen3-VL
Qwen3-VL is the multimodal large language model series developed by Qwen team, Alibaba Cloud.
File Explorer
Download Latest Version (.zip)- mobile_en_example.png
- mobile_zh_example.png
- computer_use1.jpeg
- computer_use2.jpeg
- docparsing_example1.jpg
- docparsing_example2.jpg
- docparsing_example3.jpg
- docparsing_example4.jpg
- docparsing_example5.png
- docparsing_example6.png
- docparsing_example7.jpg
- docparsing_example8.png
- chart2code_input.png
- screenshot_demo.png
- sketch2code_input.jpeg
- ocr_example1-checkpoint.jpg
- ocr_example2-checkpoint.jpg
- ocr_example3-checkpoint.jpg
- ocr_example4-checkpoint.jpg
- ocr_example1.jpg
- ocr_example2.jpg
- ocr_example3.jpg
- ocr_example4.jpg
- ocr_example5.jpg
- ocr_example6.jpg
- sample-anime-result.jpg
- sample-anime.jpeg
- sample-bird-result.jpg
- sample-bird.jpg
- sample-celebrity-2.jpg
- sample-celebrity-result.jpg
- sample-celebrity.jpeg
- sample-food-result.jpeg
- sample-food.jpeg
- hopinn.jpg
- autonomous_driving.jpg
- cam_infos.json
- dining_table.png
- drone_cars2.png
- football_field.jpg
- lots_of_cars.png
- lots_of_people.jpeg
- lounge.jpg
- manipulation.jpg
- office.jpg
- spatio_case1.jpg
- spatio_case2_aff.png
- spatio_case2_aff2.png
- spatio_case2_plan.png
- spatio_case2_plan2.png
- take_screenshot.py
- test_mmcode.py
- agent_function_call.py
- 2d_grounding.ipynb
- 3d_grounding.ipynb
- computer_use.ipynb
- document_parsing.ipynb
- german_document_ocr.ipynb
- long_document_understanding.ipynb
- mmcode.ipynb
- mobile_agent.ipynb
- ocr.ipynb
- omni_recognition.ipynb
- spatial_understanding.ipynb
- think_with_images.ipynb
- video_understanding.ipynb
- docker_web_demo.sh
- Dockerfile-qwen3vl-cu128
- common_utils.py
- dataset_utils.py
- eval_instruct.sh
- eval_think.sh
- eval_utils.py
- infer_instruct.sh
- infer_think.sh
- README.md
- requirements.txt
- run_mathv.py
- common_utils.py
- dataset_utils.py
- eval_instruct.sh
- eval_think.sh
- eval_utils.py
- infer_instruct.sh
- infer_think.sh
- README.md
- requirements.txt
- run_mmmu.py
- dataset_utils.py
- eval_instruct.sh
- eval_think.sh
- eval_utils.py
- infer_instruct.sh
- infer_think.sh
- README.md
- requirements.txt
- run_odinw.py
- common_utils.py
- dataset_utils.py
- eval_instruct.sh
- eval_think.sh
- eval_utils.py
- infer_instruct.sh
- infer_think.sh
- README.md
- requirements.txt
- run_realworldqa.py
- dataset_utils.py
- eval_instruct.sh
- eval_think.sh
- eval_utils.py
- infer_instruct.sh
- infer_think.sh
- README.md
- requirements.txt
- run_videomme.py
- 10095.png
- 10149.png
- COCO_train2014_000000580957.jpg
- v_7bUu05RIksU.mp4
- v_rBMQFpHspmo.mp4
- v_TpB_zMG3XBA.mp4
- single_images.json
- video.json
- __init__.py
- data_processor.py
- rope2d.py
- argument.py
- train_qwen.py
- trainer.py
- sft.sh
- sft_30a3b.sh
- sft_30a3b_lora.sh
- sft_32b.sh
- sft_7b.sh
- sft_qwen3_4b.sh
- zero2.json
- zero3.json
- zero3_offload.json
- check_image.py
- pack_data.py
- process_bbox.ipynb
- README.md
- __init__.py
- vision_process.py
- .python-version
- pyproject.toml
- README.md
- requirements-dev.lock
- requirements.lock
- LICENSE
- README.md
- requirements_web_demo.txt
- web_demo_mm.py
๐ Installation Guide
git clone https://github.com/QwenLM/Qwen3-VL
Downloads the entire project code from GitHub to your computer.
cd Qwen3-VL
Moves into the project folder you just downloaded.
2. Official Install Script
Easy Recommended- Python 3 Python is required to use pip.
pip install "transformers>=4.57.0"
Installs the package published on PyPI directly โ no need to clone the source.
pip install qwen-vl-utils==0.0.14
Installs the package published on PyPI directly โ no need to clone the source.
pip install -U flash-attn --no-build-isolation
Installs the package published on PyPI directly โ no need to clone the source.
pip install accelerate
Installs the package published on PyPI directly โ no need to clone the source.
Pulled directly from this repo's README.
3. Python
Easypip install "transformers>=4.57.0"
Installs the package published on PyPI directly โ no need to clone the source.
pip install qwen-vl-utils==0.0.14
Installs the package published on PyPI directly โ no need to clone the source.
pip install -U flash-attn --no-build-isolation
Installs the package published on PyPI directly โ no need to clone the source.
pip install -r requirements_web_demo.txt
Installs the Python libraries listed in requirements.txt (or similar).
pip install accelerate
Installs the package published on PyPI directly โ no need to clone the source.
Pulled directly from this repo's README.