+
+ Maintained family · Inference
+NVIDIA Cosmos Predict 2.5 generates navigable world videos. The maintained example runs the 2B text-to-world checkpoint on a single GPU.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+Black Forest Labs' FLUX family covers FLUX.1 dev and FLUX.2 (dev and distilled Klein) text-to-image. FLUX.1 registers no `model_family` in the registry and is grouped under FLUX for documentation only.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+GLM-Image from Z.ai supports both text-to-image generation and instruction-based image editing, each with a maintained example.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+HunyuanVideo 1.5 is Tencent's video generation family. The maintained examples cover 480p text-to-video with CPU offload and a full 480p-to-1080p upscale chain.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...FastVideo inference cookbook
+
+ 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.
+
+
+
+
+
-```bash
-git clone https://github.com/hao-ai-lab/FastVideo.git
-cd FastVideo
-```
+
+
+
+
+ 
+
+
+
+
-Loading…
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ + 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. +
+
+
+ Maintained family · Inference
+Kandinsky 5.0 from the Kandinsky Lab covers text-to-video and image-to-video in Lite and Pro variants, including distilled checkpoints for faster sampling.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+LongCat Video from Meituan covers text-to-video and image-to-video, and its maintained examples chain optional distilled and 720p refinement passes.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+LTX-2 from Lightricks generates video with synchronized audio. FastVideo maintains both a distilled four-GPU path and a base 1088p single-GPU path.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+Matrix Game generates playable interactive worlds from an input image and prompt. Maintained examples cover Matrix Game 2.0 variants and Matrix Game 3.0 at 720p.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Primary focus · Inference
+Generate synchronized video and audio with the full H3 checkpoint, the four-step FastH3 Preview, reference and frame-conditioned paths, or the native Apple Silicon MLX runtime.
+ 6 maintained recipes +Choose the result you want, then use a maintained CUDA or MLX path. + Device claims stay tied to checked-in sources and recorded runs.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...Maintained family · Inference
+MMAudio adds synchronized audio to video or generates audio from text. The checkpoint must be in Diffusers layout; the recipe loads the converted FastVideo repo through the `MMAUDIO_MODEL_PATH` environment variable the example reads.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+Stable Audio Open generates audio from text. FastVideo requires the converted Diffusers-format repos published by the FastVideo organization; upstream monolithic checkpoints are not loader-compatible.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...Maintained family · Inference
+Stable Diffusion 3.5 Medium is Stability AI's text-to-image model in this catalog. The maintained example sweeps a small prompt set across seeds.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...Maintained family · Inference
+TurboDiffusion profiles accelerate Wan checkpoints with step-distilled sampling and the SLA attention backend. These recipes follow the registry's `turbodiffusion` model family.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+Wan recipes span maintained CUDA examples and the released FastMetal 1.3B, 5B, and 14B native MLX paths for Apple Silicon.
+ 7 maintained recipes +Choose the result you want, then use a maintained CUDA or native MLX path.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...
+
+ Maintained family · Inference
+Z-Image Turbo from Tongyi MAI is a fast text-to-image model. The maintained example runs it on a single GPU.
+Choose a model, then pick a GPU count. Each selection maps to one + checked-in recipe. If that recipe does not use your chosen count, the + page says so instead of guessing.
+Loading recipe details...
+Exact device and memory details appear only when a recorded run supports them.
+ +Loading...