CogVideoX models

3 CogVideoX models from Z.ai on Hugging Face, from 1.7B to 5.6B parameters, published by Z.ai in F16 and BF16. The smallest official model, CogVideoX-2b, needs about 3.8 GB of VRAM at its published precision; the cheapest live fit is RTX 3070 at $0.088/hr.

Pick a size

One row per official CogVideoX size: the VRAM it needs at each precision, the cheapest GPU that holds it today, and the KV cache for a 32K-token context.

ModelParametersNative VRAMFP8 VRAMINT4 VRAMLive GPU fit (native)Est. $/hrKV cache at 32K
CogVideoX-2b1.7B3.8 GB1.9 GB0.9 GBRTX 3070$0.088/hrnot published for this architecture
CogVideoX-5b5.6B12.5 GB6.2 GB3.1 GBRTX A4000$0.167/hrnot published for this architecture

VRAM is weights times a flat 1.2 overhead; the KV cache is a separate per-model figure at 16-bit, shown where the architecture is published. See the methodology.

Official models (3)

ModelParametersPublished asVRAM neededLive GPU fitEst. $/hr
CogVideoX-2b1.7BF163.8 GBRTX 3070$0.088/hr
CogVideoX-5b5.6BBF1612.5 GBRTX A4000$0.167/hr
CogVideoX-5b-I2V5.6BBF1612.6 GBRTX A4000$0.167/hr

VRAM is for the precision the model is published in: the weights times a flat 1.2 overhead, with the KV cache not included. See the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.

More from Z.ai

Start ComfyUI on a rented GPU.

© 2026 Aquanode. All rights reserved.

All trademarks, logos and brand names are the property of their respective owners.