Skip to content
localmodel.run

video model · cogvideox · Linux

Can I run CogVideoX-5B on Nvidia DGX Spark (128GB)?

Compatibility verdict VRAM check
Yes, it runs fast on this GPU

Yes. CogVideoX-5B runs on Nvidia DGX Spark (128GB) at INT8 / fp8 (~16 GB of ~112 GB usable).

Needs ~16 GB Device usable ~112 GB

Runs at INT8 / fp8 using ~16 GB of ~112 GB usable.

Peak VRAM
~16 GB
Usable on device
~112 GB
Device memory
128 GB
Quant
INT8 / fp8
Share on X Share on Reddit

How to run it

Use Diffusers or ComfyUI at INT8 / fp8. The big text encoder is loaded to encode your prompt, then offloaded before generation, which is why peak VRAM stays near the backbone size rather than the sum of every file.

Model cogvideox
Type
video (DIT)
Parameters
5B
Peak VRAM
~16 GB at INT8 / fp8
Resolution
720×480
License
CogVideoX License
Full CogVideoX-5B requirements →
Device Linux
Memory
128 GB unified
Usable for weights
~112 GB
Power draw
~140 W
Best runtime
llama.cpp (CUDA) / Ollama (CUDA)
Best models for Nvidia DGX Spark (128GB) →

You could also run

Run CogVideoX-5B on other hardware

Nvidia GeForce RTX 4090 (24GB)Nvidia GeForce RTX 3090 (24GB)AMD Radeon RX 7900 XTX (24GB)Nvidia GeForce RTX 5090 (32GB)AMD Ryzen AI Halo (128GB)

FAQ

Can Nvidia DGX Spark (128GB) run CogVideoX-5B?

Yes. CogVideoX-5B runs on Nvidia DGX Spark (128GB) at INT8 / fp8 (~16 GB of ~112 GB usable).

How much VRAM does CogVideoX-5B need?

Nvidia DGX Spark (128GB) has room to spare. At INT8 / fp8 the realistic peak is ~16 GB of VRAM, versus ~26 GB with every component kept resident (no offload). With aggressive CPU offload it drops to ~5 GB, much slower.

What do I use to run CogVideoX-5B locally?

CogVideoX-5B runs in Diffusers or ComfyUI. It loads as a video diffusion checkpoint plus its text encoder and VAE, not a single chat command.

Sources

VRAM figures are sourced peak-usage anchors at the noted quant; catalog updated 2026-11-01. See methodology.