How to run CogVideoX-5B locally

Zhipu AI / THUDM · Text → video · 5B params · CogVideoX License (custom) (non-commercial)

The 5B flagship of the CogVideoX family, an early high-quality open T2V model that runs on modest GPUs but ships under a custom (non-Apache) license.

What it costs to run — $0 markup

• On your own machine — $0. Runs free if you have about 5.0 GB of VRAM (from ~5GB with diffusers BF16 + CPU offload/quant; ~15GB unoptimized peak), via ComfyUI / Diffusers.
• Rent a GPU — from $0.26/hr. Fits RTX 3090 24GB at the direct vendor price ($0 markup) — pay only for the minutes you generate.
• Download the weights — free. Open weights at THUDM/CogVideoX-5b.

Note: generative-media models are billed per image / per second / per minute on hosted services — not per token. Running locally or on your own rented GPU is usually far cheaper and keeps your data on your machine.

Key facts

Get CogVideoX-5B on Hugging Face →

More video models

• Wan2.2 T2V-A14B — Text → video, ~24 GB VRAM
• Wan2.2 I2V-A14B — Image → video, ~24 GB VRAM
• HunyuanVideo — Text → video, ~45 GB VRAM
• Wan2.2 TI2V-5B — Text → video, ~8.0 GB VRAM
• LTX-Video (13B) — Text → video, ~13 GB VRAM
• Stable Video Diffusion (img2vid-XT) — Image → video, ~16 GB VRAM

Browse: all media models · chat / LLM models

Open the free Spanvero advisor → · We point you to the open weights + your own accounts, $0 markup, never resell compute. © 2026 Cynosure LLC.