Zhipu Qingying 智谱清影 (CogVideoX)
Zhipu's AI video generator, from the Qingyan app to a ¥1-a-clip API
Overview
Qingying (清影) is Zhipu AI's text-to-video and image-to-video feature, launched in July 2024 inside the Zhipu Qingyan app (chatglm.cn) and offered to developers through the bigmodel.cn API. The hosted model is now CogVideoX-3 (5- or 10-second clips up to 4K, ¥1 per video); Zhipu's earlier CogVideoX models, built with Tsinghua University researchers, are open-weight on GitHub.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 12 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
What you can produce with Zhipu Qingying 智谱清影 (CogVideoX)
- Text-to-video and image-to-video in the Zhipu Qingyan app (chatglm.cn)
- CogVideoX-3 API clips of 5 or 10 seconds up to 4K, with first- and last-frame control
- The free CogVideoX-Flash API model for prototyping
- Open CogVideoX weights to self-host (2B Apache 2.0; 5B and 1.5-5B under the CogVideoX License)
- LoRA fine-tuning of open CogVideoX models through CogKit or diffusers
- Local CogVideoX1.5-5B inference in diffusers from about 10GB of GPU memory with optimisations
Qingying is a cheap, capable hosted video generator: CogVideoX-3 makes 5- or 10-second clips up to 4K for ¥1 each through the API, and a free CogVideoX-Flash model suits prototyping. The open-weight CogVideoX line (2B under Apache 2.0, 5B under a stricter licence) remains a useful research and fine-tuning base, but it has not been updated since late 2024 and the current hosted model is not open.
For ASEAN users, access is the hurdle: the consumer app is Mandarin-first and needs a Chinese phone or WeChat login, and billing is in yuan. Best fit: developers already using Zhipu's platform who want low-cost video generation.
What we found
Qingying (清影) is the video-generation feature Zhipu AI launched on 26 July 2024 inside its Zhipu Qingyan assistant, with the API released the same day on Zhipu's bigmodel.cn platform. Chinese tech outlet ifanr noted it opened to everyone on day one with no waitlist. Zhipu, which listed on the Hong Kong exchange in January 2026, still links to "QingYing" from its CogVideoX repository, but the logged-out chatglm.cn sidebar now simply says "AI生视频" (AI video), and generation sits behind a phone or WeChat login.
The model underneath is CogVideoX, built with Tsinghua University researchers. Its open-weight line stopped at CogVideoX1.5-5B in November 2024: the 2B model is Apache 2.0, while the 5B models use a separate, more restrictive CogVideoX licence, and the repository's most recent commit dates from November 2025. Zhipu's own README lists the 1.5-5B model running in diffusers from about 10GB of GPU memory with optimisations, which keeps local experiments within reach of a single consumer card.
The hosted model has moved on and is not open. CogVideoX-3 adds first-and-last-frame generation, 5- or 10-second clips and resolutions up to 4K, at a flat ¥1 per video through the API; an earlier version of this page said Zhipu published no per-video rate. CogVideoX-Flash is free, and the older CogVideoX-2 costs ¥0.5. The same price list also sells third-party Vidu video models.
For ASEAN users the obstacles are access rather than quality: the consumer product is Mandarin-first and requires a Chinese login, and the API is billed in yuan.
The honest positioning is narrower than the old one. Qingying is a cheap, capable hosted generator for people already in Zhipu's ecosystem, and CogVideoX remains a useful open research base for fine-tuning. It is not the frontier of video generation, and its open weights have not been refreshed since 2024.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
About this listing
This entry was compiled from publicly available data including Zhipu Qingying 智谱清影 (CogVideoX)'s official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Zhipu Qingying 智谱清影 (CogVideoX) unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Zhipu Qingying 智谱清影 (CogVideoX) directly →
Spotted something out of date? Suggest an update →
Related reading
Alternatives to Zhipu Qingying 智谱清影 (CogVideoX)
More in Video & Audio
