dgx-spark-qwen38
GitHub

dgx-spark-qwen38 / Models / Qwen-Image 2.1

Run Qwen-Image 2.1 on a DGX Spark

Images from a prompt, by Qwen. On SGLang.

Qwen's image model on SGLang. Its Turbo, Qwen's eight-step distillation of the same model, is a second checkpoint of the lane. On an ASUS Ascent GX10 with 128 GB: 7.5 s, with its Turbo, per 1024×1024 image.

Measured

7.5s

with its Turbo, per 1024×1024 image

  • 34.3 s at 40 steps without it
  • The Turbo fetches 32.5 GB at its first switch
  • On the same box as the text lanes
Measured on
an ASUS Ascent GX10 with 128 GB, the project's reference box
How
1024×1024, timed on the box
Source
docs/image-lane.md

Load it

From the cockpit: open the Lanes view, pick Qwen-Image 2.1 and press Load. From a terminal, in the repo folder on the box:

./prepare-lane.sh image   # the first time only, builds the lane
./switch-model.sh image   # switches the boot, then prints the commands to run

switch-model.sh never restarts an engine itself: it prints the exact stop and start commands. The cockpit's Load runs them for you.

Other checkpoints of the same lane load the same way: image-turbo.

On a 64 GB DGX Spark

No: it needs a 128 GB DGX Spark. A 64 GB box serves Nemotron 3.5 Bolt and Qwen3.8-27B: what runs on 64 GB.

Use it

From the cockpit: its Image view drives the lane. The lane listens on 127.0.0.1:30020, on the box itself, and not through the API on port 30001.

Install it

One command, on stock DGX OS. The same command updates it later.

curl -fsSL https://raw.githubusercontent.com/hasso5703/dgx-spark-qwen38/main/get.sh | bash

Questions about Qwen-Image 2.1 on a DGX Spark

Can a DGX Spark run Qwen-Image 2.1?

Yes. Qwen's image model on SGLang. Its Turbo, Qwen's eight-step distillation of the same model, is a second checkpoint of the lane. The installer of dgx-spark-qwen38 sets it up, and the cockpit loads it with one button.

How fast is Qwen-Image 2.1 on a DGX Spark?

7.5 s, with its Turbo, per 1024×1024 image, measured on an ASUS Ascent GX10 with 128 GB. 34.3 s at 40 steps without it. The Turbo fetches 32.5 GB at its first switch. On the same box as the text lanes.

Does Qwen-Image 2.1 run on a 64 GB DGX Spark?

No: it needs a 128 GB DGX Spark.

The other lanes