--- hide: - toc --- # Inference Cookbook

FastVideo inference cookbook

Choose a model family.

Open a family to pick a maintained recipe and a runtime FastVideo actually supports. Every command runs a checked-in source, so the model, platform, offload, and attention settings stay tied to that example. The catalog is derived from the model families registered in fastvideo/registry.py.

View the full support matrix
MiniMax H3Video + audio · CUDA + MLX 6 recipes WanVideo generation · CUDA + MLX 7 recipes LTXVideo and audio generation 2 recipes HunyuanVideo generation 2 recipes CosmosWorld and video generation 1 recipe Kandinsky 5Text and image to video 2 recipes FLUXImage generation 3 recipes GLM-ImageImage generation and editing 2 recipes Z-ImageImage generation 1 recipe Stable DiffusionImage generation 1 recipe LongCatVideo generation 2 recipes Stable AudioAudio generation 2 recipes MMAudioAudio generation 1 recipe Matrix GameInteractive world generation 2 recipes TurboDiffusionAccelerated Wan profiles 3 recipes
GameCraftGame world generation Page planned
GEN3CNovel-view video Page planned
HY-WorldInteractive world play Page planned
DreamXWorld generation Page planned
LingBotVideo and world models Page planned

Inference first, then the full workflow

Inference is the first complete stage. Distillation, fine-tuning, training, evaluation, optimization, and deployment will reuse the same family-first structure as their recipes land. Each family page shows which stages are available and which are planned.

Catalog marks come from the official model publishers' Hugging Face organizations; typographic tiles are placeholders, never invented logos. See docs/assets/logos/SOURCES.md.