Skip to main content
Rase AI Intelligence Newsletter

Creative AI Model or AI service

CogVideoX expanded from text-to-video to image-to-video

CogVideoX is a family of downloadable video generation models. Its original text-to-video checkpoints appeared in August 2024, followed by an image-to-video version in September. [1][2]

Model access
Original 2B and 5B text-to-video weights and separate 5B image-to-video weights are available. [1]
License
Apache 2.0 for the 2B model and repository code. Separate CogVideoX terms cover the original 5B models. [1]

What the model does

The project released the 2B model on August 6 and the larger 5B on August 27. On September 19 it released a distinct 5B image-to-video model that accepts a starting image and a text prompt. [1]

Why it matters

A designer starting with an image can use a model that conditions on the original frame. A team with only a written brief can use the text-to-video weights. [1][2]

Who can use it

People testing creative video workflows can compare the model sizes and input modes. Checkpoint choice affects hardware requirements and which type of input is supported. [1][2]

Key dates

  1. CogVideoX-2B released [1]
  2. CogVideoX-5B text-to-video released [1]
  3. CogVideoX-5B image-to-video released [1]

Ways to use it

For a storyboard test, compare whether several generated clips follow the same simple prompt while preserving the important subjects. Use the separate image-to-video checkpoint when a starting image matters. [1][2]

Model limitations

The 2B and 5B models have different weight licenses. Supported clip lengths, resolutions and frame behavior depend on the exact version. [1][2]

Sources and verification

Catalog evidence checked How we verify records

  1. Primary source Official CogVideoX checkpoint release history and license distinctions
  2. Primary source THUDM/CogVideoX-5b
  3. Other source CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer

Related profiles

Creative AI

Seedance 2.5 (4K API and Flash)

ByteDance (Volcano Engine / BytePlus)

These tools generate video through developer services. The cited sources describe different variants and routes of access.

Medium confidenceEvidence checked

Creative AI

LongCat-Video-Avatar 1.6

Meituan (LongCat team)

It generates video of an avatar from supported inputs in the documented project.

High confidenceEvidence checked

Creative AI

MiniMax H3-Image

MiniMax

It generates images from prompts through the model architecture described by MiniMax.

High confidenceEvidence checked