FastWan2.2-TI2V-5B โ€” MLX (DMD 3-step)

Self-contained MLX conversion of FastVideo/FastWan2.2-TI2V-5B-FullAttn for Apple-Silicon image-to-video, used as the Fast tier in Videoboom. DMD-distilled: 3 denoise steps (sigmas [1.0, 0.757, 0.522, 0.0], guide=1, renoise) via mlx-video.

Files (self-contained)

file size what
model.safetensors 10.0 GB DiT (bf16)
t5_encoder.safetensors 11.4 GB umt5 text encoder
vae.safetensors 2.8 GB VAE
config.json โ€” includes the fastwan_dmd marker

~832ร—480 / 121 frames per clip. License Apache-2.0 (inherited from the base model).

Downloads last month
55
Safetensors
Model size
5B params
Tensor type
BF16
ยท
MLX
Hardware compatibility
Log In to add your hardware

Quantized

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for lBroth/FastWan2.2-TI2V-5B-MLX

Finetuned
(1)
this model