What CogVideoX does
CogVideoX is an open-source text and image to video model family from the CogVideo repository, licensed under Apache-2.0. The README says it helps developers get started with basic usage and fine-tuning. Listed models include CogVideoX1.5-5B, CogVideoX1.5-5B-I2V, CogVideoX-2B, and CogVideoX-5B with its image-to-video variant. Supported tasks are text-to-video, image-to-video, and video continuation. Other features include a prompt conversion script that uses an LLM such as GLM-4 to expand prompts, INT8 and FP8 quantized inference for lower-memory hardware, and LoRA and SAT-based fine-tuning. Models are hosted on Hugging Face and ModelScope, and the repository lists Colab notebooks and a Gradio demo. The page does not state pricing for the open-source project.
How CogVideoX compares with its closest alternatives
All 5 closest alternatives have a free plan, as CogVideoX does. All 5 offer an API, as CogVideoX does. CogVideoX offers a desktop app. Of the alternatives, 1 offers a desktop app. CogVideoX was founded in 2024; the alternatives were founded between 2023 and 2024, 2 of them earlier. CogVideoX has no single headquarters on record; its alternatives with a known headquarters are in the United States and Israel.