AIVideo combines generative media models, a native editor, workflows, and automations for creating production content.
Pyramid Flow
Pyramid Flow explores pyramidal flow matching for efficient generative video modeling.
About Pyramid Flow
Pyramid Flow is a research project page for an autoregressive video generation model based on pyramidal flow matching. The work is presented by researchers from Peking University, Kuaishou Technology, and the Beijing University of Posts and Telecommunications. The page shows sample text-to-video and image-to-video clips and states that the model was trained only on open-source datasets using 20.7k A100 GPU hours.
The page links to a paper, code, a model, and a demo. The excerpt does not give scores for its quantitative results, which are listed only by benchmark name (VBench and user preference). Pricing is unknown, and the excerpt does not state terms for public use.
Key features
- Text-to-video generation at 1280x768 resolution, 24fps, with 5-second and 10-second clips
- Text-conditioned image-to-video generation at 1280x768 resolution, 24fps, 5-second clips
- Autoregressive video generation based on flow matching
- Training stated to use only open-source datasets
- Reported training cost of 20.7k A100 GPU hours
Best for: Researchers and developers who want to study efficient autoregressive video generation methods based on flow matching.
Good for
- States that training used only open-source datasets
- Reports the training compute cost in A100 GPU hours
- Shows example outputs for both text-to-video and image-to-video generation
Watch out for
- Quantitative results are named (VBench, user preference) but no values appear in the excerpt
- Pricing is unknown
Pyramid Flow at a glance
- Website
- pyramid-flow.github.io
- Category
- AI & Machine Learning › AI Video Generators
- Type
- Website
- Pricing
- Not confirmed. Check the site
- Platforms
- Web
- Open source
- Not stated
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
Pyramid Flow alternatives
Tools and sites you could use instead, ranked by how closely they match.
Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models.
Fllow.ai generates videos using models including Sora, Veo, Hailuo, Wan, Kling, and Seedance through one video platform.
Pika provides video and image generation tools, with model versions and templates listed on its creative platform.
Genmo develops video generation models and offers video creation with Mochi.
Open source
24 alternatives
fal provides an API for integrating the Seedance 2.0 video generation model.
Popular in AI Video Generators
Well-known listings from the same topic.
AIFreeVideoaifreevideo.comGstory.aigstory.aiDubprodubpro.aiVibesvibes.aiPlask Motionplask.aiFaceless Reelsfacelessreels.app
Pyramid Flow FAQ
What training data was used for the model?
The page states that the model was trained only on open-source datasets.
What video resolution and length are shown in the examples?
The text-to-video examples are 1280x768 at 24fps, in 5-second and 10-second lengths. The image-to-video examples are 1280x768 at 24fps, 5 seconds long.
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.