Qwen3.8-Flash-Next-FP4-Local: Answer questions about images and video (125B)
Added · Updated
Made by NVIDIA
View profile on Hugging Face (opens in a new tab)Ask questions about text, images, or video and receive text responses.
For teams with supported NVIDIA Blackwell hardware and the documented vLLM containers.
Model details
Original model nvidia/Qwen3.8-Flash-Next-FP4-Local
Input Image or text Output Text
Ways to use it Run locally
Role in this family: Smaller quantized version.
Qwen/Qwen3.8-Flash-Next (opens in a new tab)What it does
This NVIDIA checkpoint is a quantized version of Qwen3.8-Flash-Next that accepts text, images, and video and returns text. The model card reports 125B total and 6B active parameters, plus a separate n-gram embedding table. The checkpoint occupies about 96 GiB on disk. The card lists local vLLM serving on NVIDIA Blackwell hardware, with validation on an RTX PRO 6000 Blackwell and DGX Spark. It points users to NVIDIA’s Open Model Agreement for this checkpoint and the Qwen Community License for the base model.
How to use it
Download the checkpoint under the listed NVIDIA Open Model Agreement terms, start one of the containers linked in the card, and launch the provided vLLM command. Send text, image, or video input through the running model interface and read its text response.
Common use cases
- Send a picture with a question about its contents and receive a written answer.
- Provide a video and ask for a text description of what is shown.
- Connect the served model to an application that needs text responses to mixed text, image, or video inputs.
Can I use this?
Requirements The card specifies Linux, NVIDIA Blackwell hardware, vLLM, and provided community containers. It reports validation on an RTX PRO 6000 Blackwell and DGX Spark, but does not state a universal minimum memory requirement. The download is about 96 GiB. NVIDIA Open Model Agreement terms govern this checkpoint; the Qwen base model has a separate Qwen Community License.
Caveat The new NVFP4 scale-factor format is not yet supported by upstream vLLM according to the card; use one of the provided containers. Review both the NVIDIA agreement and the Qwen base-model license.
Worth knowing The model card says the NVIDIA Open Model Agreement governs this checkpoint and names a separate Qwen Community License for the base model; review both linked terms. NVIDIA’s agreement permits commercial use and says NVIDIA does not claim ownership of generated outputs. The card says upstream vLLM support for its scale format has not landed, so use a provided container.
Why it's here
Not trending this week
Last seen on the Hugging Face trending list on 22 October 2026. It stays on BuildTube so you can still find it.
- 394 people have liked it on Hugging Face.
- It was downloaded 34,115 times in the last 30 days.
Numbers checked on 8 November 2026; not refreshed since.
Behind it
View model pageBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?
Quick questions
- What is Qwen3.8-Flash-Next-FP4-Local?
- This NVIDIA checkpoint is a quantized version of Qwen3.8-Flash-Next that accepts text, images, and video and returns text.
- What can I use Qwen3.8-Flash-Next-FP4-Local for?
- Send a picture with a question about its contents and receive a written answer; Provide a video and ask for a text description of what is shown; Connect the served model to an application that needs text responses to mixed text, image, or video inputs
- Can I run Qwen3.8-Flash-Next-FP4-Local on my computer?
- At about 125 billion parameters in total (6 billion active), it needs server- or workstation-class GPU hardware, not an ordinary laptop or desktop. The card specifies Linux, NVIDIA Blackwell hardware, vLLM, and provided community containers. It reports validation on an RTX PRO 6000 Blackwell and DGX Spark, but does not state a universal minimum memory requirement.
- What do I need to use Qwen3.8-Flash-Next-FP4-Local?
- Linux, supported NVIDIA Blackwell hardware, and one of the provided vLLM containers; the checkpoint occupies about 96 GiB on disk.
- What license does Qwen3.8-Flash-Next-FP4-Local use?
- The model card says the NVIDIA Open Model Agreement governs this checkpoint and names a separate Qwen Community License for the base model; review both linked terms. Review both the NVIDIA agreement and the Qwen base-model license.
More from this week
- NVKE Run managed Kubernetes clusters on your own NVIDIA GPU machines
Set up a service that creates and manages separate Kubernetes clusters on your own NVIDIA GPU servers.
- Idol Turn one picture of an object into a 3D Gaussian splat
Idol turns one photo, render or generated image of an object into a 3D Gaussian splat file.
- P-H3-Image-Comfy Make and edit 8-megapixel images in ComfyUI
Turn short text requests into 8-megapixel images and edit pictures on your own graphics card.
Explore by topic
Post this
Download share imageFor makers
Made this? Show it’s on BuildTube — add the badge to your README or embed the card on your site.
README badge (Markdown)
[](https://buildtube.app/p/hf-nvidia-qwen3-8-flash-next-fp4-local/)Embed card (HTML)
<iframe src="https://buildtube.app/embed/hf-nvidia-qwen3-8-flash-next-fp4-local/" width="560" height="240" style="border:0;max-width:100%" loading="lazy" title="Qwen3.8-Flash-Next-FP4-Local on BuildTube"></iframe>