CogVideoX-5B
Zhipu's 5B open text-to-video model — 10s clips, the first commercial-grade open video generator.
About
CogVideoX-5B, released August 27, 2024 by Zhipu AI with Tsinghua University's KEG lab, was among the first commercial-grade open-source text-to-video models. Its expert-transformer architecture with a 3D causal VAE generates coherent 6–10 second clips at up to 768x1360, and the 5B release includes both text-to-video and image-to-video checkpoints (the 1.5 refresh followed in November 2024).
Licensing nuance: the code is Apache 2.0 and the 2B model is Apache 2.0, but CogVideoX-5B itself ships under a custom CogVideoX license — read it before commercial deployment. Inference is heavy (26GB+ VRAM in full precision, ~5GB minimum with diffusers INT8). Newer open video models have since surpassed it, but it remains a well-documented baseline for video-generation research.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 65; Trending score 70/100 from researcher assessment.