This model requires you to accept the Apache 2.0 license terms before download.
Description
Qwen3-VL 30B-A3B Thinking — text, image and video understanding; 128-expert MoE, DeepStack vision, interleaved M-RoPE; multi-GPU sharded
Tags
At a glance
| Kind | multimodal model |
|---|---|
| Precision | bfloat16 |
| Container size | 57.9 GiB |
| Weights inside | 57.9 GiB |
| Components | 3 |
| Licence | Apache 2.0 |
What is inside the container
A .nbx container packs every part of Qwen3-VL-30B-A3B-Thinking — 3 components with weights, graphs and runtime settings — so one file is all the runtime needs.
| Component | Blocks | Layers | Weights |
|---|---|---|---|
| model.language_model | 0 | 48 | 56.3 GiB |
| model.visual | 0 | — | 1.0 GiB |
| lm_head | 0 | — | 594 MiB |
Default settings
The values the container ships with. They apply when you do not pass your own.
- Max length
- 262144
- Temperature
- 0.6
- Top-p
- 0.95
- Top-k
- 20
Memory needed
The weights of Qwen3-VL-30B-A3B-Thinking take 57.9 GiB in bfloat16. That is the floor for the memory the runtime has to place, spread over the devices it finds; activations come on top and grow with the size and length you ask for. A measured peak per device class is not published for this model yet.
Verify what you downloaded
Every download carries its SHA-256, and the file endpoint sends it in the X-NBX-SHA256 header. The container is 57.9 GiB (62,146,134,424 bytes).
$ sha256sum Qwen3-VL-30B-A3B-Thinking.nbx
e503a61f45fb525f4780991de860b23bfa65195a64ef253a3f065a2669f77646
Questions
What is Qwen3-VL-30B-A3B-Thinking?
A multimodal model by Qwen, distributed as a bfloat16 .nbx container and licensed Apache 2.0.
How big is the download?
57.9 GiB, of which 57.9 GiB are weights.
How do I run it?
Install the runtime from the download page, then use the commands under “CLI Usage” below: neurobrix import Qwen/Qwen3-VL-30B-A3B-Thinking, then neurobrix serve --model Qwen3-VL-30B-A3B-Thinking.
CLI Usage
$ neurobrix import Qwen/Qwen3-VL-30B-A3B-Thinking
$ neurobrix serve --model Qwen3-VL-30B-A3B-Thinking
$ neurobrix run --prompt "your prompt here"
$ neurobrix stop