Aenica Video Index
← Back to models

Z.ai

CogVideoX1.5-5B

Active

CogVideoX1.5-5B is Z.ai’s provider-published local family with official Diffusers text-to-video and image-to-video checkpoints plus a SAT weights repository.

Capabilities

Text-to-videoVaries by variantImage-to-videoVaries by variantVideo-to-videoUnknownVideo editingNo

Local deployment evidence

The official repository provides SAT and Diffusers local inference paths and identifies CogVideoX1.5-5B as the latest supported family.

Hardware entry summary

Lowest documented entry VRAM
9 GB
Applies to
CogVideoX1.5-5B I2V (Diffusers)

The 9 GB entry value applies only to the optimized Diffusers I2V workflow and is not a universal family minimum.

View full hardware profiles ↓

Local checkpoints

CogVideoX1.5-5B T2V (Diffusers)

Variant ID: cogvideox1-5-5b-t2v-diffusers

Active
Text-to-videoYesImage-to-videoNoVideo-to-videoUnknownVideo editingNo
Access mode
Local checkpoint
Checkpoint type
Full
Precision
BF16
Maximum duration
10 seconds
Maximum resolution
1360 × 768
Frame rates
16 fps
Aspect ratios
Not specified
Native audio
No
Reference image
No
Pricing
No provider inference fee
View official weights page

Notes

Official local Diffusers checkpoint. There is no provider inference fee, but hardware, electricity, storage, and processing time still cost money. Commercial use requires compliance with the CogVideoX License.

CogVideoX1.5-5B I2V (Diffusers)

Variant ID: cogvideox1-5-5b-i2v-diffusers

Active
Text-to-videoNoImage-to-videoYesVideo-to-videoUnknownVideo editingNo
Access mode
Local checkpoint
Checkpoint type
Full
Precision
BF16
Maximum duration
10 seconds
Maximum resolution
768–1360 px, dimensions divisible by 16
Frame rates
16 fps
Aspect ratios
Not specified
Native audio
No
Reference image
Yes
Pricing
No provider inference fee
View official weights page

Notes

Official local Diffusers checkpoint. There is no provider inference fee, but hardware, electricity, storage, and processing time still cost money. Commercial use requires compliance with the CogVideoX License.

Hardware profiles

Optimized Diffusers T2V

Applies to
CogVideoX1.5-5B T2V (Diffusers)
Official minimum VRAM
10 GB
Official recommended VRAM
Unknown
Official minimum system RAM
Unknown
Official recommended system RAM
Unknown
Storage
Unknown

The official model card reports Diffusers BF16 from 10 GB with its documented optimizations enabled; testing was performed on A100/H100.

Official documentation

Optimized Diffusers I2V

Applies to
CogVideoX1.5-5B I2V (Diffusers)
Official minimum VRAM
9 GB
Official recommended VRAM
Unknown
Official minimum system RAM
Unknown
Official recommended system RAM
Unknown
Storage
Unknown

The official I2V model card reports a 9 GB BF16 minimum with all listed Diffusers optimizations enabled and cautions that testing used A100/H100.

Official documentation

Overview

Provider
Z.ai
Status
Active
Publication status
Published
Record scope
Family
Release date
2024-11-08
Last verified
2026-08-06

Specifications

Family maximum duration
10 seconds
Family maximum resolution
Varies by variant
Frame rates
16 fps
Aspect ratios
Not specified
Portrait support
Unknown
Native audio
No
Reference image
Varies by variant
Character consistency
Unknown

Hardware

Local deployment
Yes
Tested minimum VRAM
Unknown
ComfyUI support
Unknown

Licensing

Open weights
Yes
Open source
No
Commercial use
Conditional
License
CogVideoX License

Availability

API availability
Unknown
Web app availability
Unknown

Performance and pricing

Speed category
Unknown
Estimated clip cost
Unknown
Pricing
Free

Best use cases

  • Local 10-second text-to-video generation
  • Local image-to-video generation with flexible dimensions

Limitations

  • Official low-VRAM figures depend on aggressive Diffusers optimizations and A100/H100 testing.
  • The official ComfyUI link is listed among community adaptations, so support is not marked true.

Hardware notes

The official T2V and I2V pages give different optimized Diffusers entry figures. They remain separate profiles.

Commercial use notes

Commercial use is conditional. The CogVideoX License requires registration for a basic commercial license and further licensing above one million monthly service visits.

Speed notes

Official hardware-specific timings are not converted into an Aenica speed category; no reproducible local test has been completed.

Pricing notes

Local checkpoints have no provider inference fee but require user-funded compute, storage, electricity, and time.

Sources