Skip to content
ModelListedExternal
Browse files

Wan 2.2 T2V A14B Diffusers BF16

This repository contains BF16 MLX-Gen saved weights for Wan-AI/Wan2.2-T2V-A14B-Diffusers. It is designed for local Apple Silicon inference with mlx-gen.

Sourcehuggingface.co/AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16/tree/2061f003591d4b3377afab7dc6d3aaea92d1107b
KindOtherBase modelWan-AI/Wan2.2-T2V-A14B-DiffusersVersionRevision 2061f00Unratedgrade

Model source

Kind
Other
File size
242 MB–4.64 GB
Base model
Wan-AI/Wan2.2-T2V-A14B-Diffusers
Version
vRevision 2061f00
Updated
Nov 3, 2026
Downloads
0
Source
Hugging Face

Source description

wan2.2-t2v-a14b-diffusers-bf16

This repository contains BF16 MLX-Gen saved weights for Wan-AI/Wan2.2-T2V-A14B-Diffusers. It is designed for local Apple Silicon inference with mlx-gen.

It uses the mflux/MLX saved-weight layout. It is not a Diffusers or Transformers from_pretrained() checkpoint.

Source Model

Original model: Wan-AI/Wan2.2-T2V-A14B-Diffusers.

This prepared derivative follows the Apache 2.0 license of the source model.

Precision

This package stores the Wan A14B T2V transformer and VAE weights for MLX-Gen BF16 runtime use. The UMT5 text encoder, scheduler metadata, tokenizer files, and model index are included in the prepared folder.

Validation

Measured on 2026-06-04 with mlx-gen 0.18.9 on Apple Silicon. The upstream Diffusers source snapshot measured about 118 GiB in the local Hugging Face cache before preparing these packages. The table below reports prepared-package generation from model init through MP4 save and post-save video-health validation.

Validation profile: 384x224, 33 frames, 12 denoising steps, guidance 4, guidance-2 3, 8 fps, seed 4242, --low-ram.

PackageDiskFull-Process Physical PeakMax RSSMLX PeakTotal TimeVideo Health
This BF16 package64.3 GiB33.0 GiB31.8 GiB27.7 GiB152.7 s33/33 frames, 384x224, 8 fps, temporal delta 1.3
Mixed q8/BF16 package39.7 GiB20.7 GiB19.5 GiB15.5 GiB154.8 s33/33 frames, 384x224, 8 fps, temporal delta 1.4

Physical peak is Darwin ri_phys_footprint sampled for the full process. The validation is intentionally small and repeatable; it is not a claim that every full-size 1280x720, 81-frame, 40-step job has the same memory or timing profile.

Usage

python -m pip install -U mlx-gen

mlxgen download --model AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16

mlxgen generate \
  --model AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16 \
  --task text-to-video \
  --prompt "A cinematic scene of a scientist working on agentic AI through the night, monitors glowing, papers shifting in a slow dolly shot." \
  --width 384 \
  --height 224 \
  --frames 33 \
  --steps 12 \
  --guidance 4 \
  --guidance-2 3 \
  --fps 8 \
  --seed 4242 \
  --low-ram \
  --metadata \
  --output video.mp4

Compatibility

Requires mlx-gen >= 0.18.9.

Generated with mlx-gen 0.18.9.

Use the mlxgen command and Python import path for new MLX-Gen projects.

Attribution

MLX-Gen is based on mflux by Filip Strand and the original mflux contributors.

Prepared and contributed by @lpalbou.

Sources

1 source
Hugging FacePrimaryVersion 2061f003591d4b3377afab7dc6d3aaea92d1107b
huggingface.co/AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16/tree/2061f003591d4b3377afab7dc6d3aaea92d1107b

Verified Nov 3

Download files

text_encoder · 3 weight files

All files listed in the provider index are present.

transformer · 14 weight files

All files listed in the provider index are present.

transformer_2 · 14 weight files

All files listed in the provider index are present.

vae · 1 weight files

All files listed in the provider index are present.

Repository files

File inventory only; completeness is unverified.

Model artifacts

32 artifacts

text_encoder/model-00001-of-00003.safetensors

safetensors · 4.60 GB · SHA-256 a8e861969c74…8f58 · Hugging Face

Download shard

text_encoder/model-00002-of-00003.safetensors

safetensors · 4.64 GB · SHA-256 d57d948ece48…02d7 · Hugging Face

Download shard

text_encoder/model-00003-of-00003.safetensors

safetensors · 1.34 GB · SHA-256 0da9ee284e21…7c5f · Hugging Face

Download shard

transformer_2/0.safetensors

safetensors · 1.99 GB · SHA-256 f9ee806421df…7e92 · Hugging Face

Download file

transformer_2/1.safetensors

safetensors · 1.96 GB · SHA-256 eb55633aaeb2…f158 · Hugging Face

Download file

transformer_2/10.safetensors

safetensors · 1.96 GB · SHA-256 94b781cf8b59…c5ab · Hugging Face

Download file

transformer_2/11.safetensors

safetensors · 1.96 GB · SHA-256 f437bf829039…7013 · Hugging Face

Download file

transformer_2/12.safetensors

safetensors · 1.96 GB · SHA-256 8fe482282a1d…306a · Hugging Face

Download file

transformer_2/13.safetensors

safetensors · 1.07 GB · SHA-256 1b45445e9600…5700 · Hugging Face

Download file

transformer_2/2.safetensors

safetensors · 1.96 GB · SHA-256 e11c8e7cadae…0bcb · Hugging Face

Download file

transformer_2/3.safetensors

safetensors · 1.96 GB · SHA-256 1abc8a6e026c…e807 · Hugging Face

Download file

transformer_2/4.safetensors

safetensors · 1.96 GB · SHA-256 d5a43ede6ddb…b5c0 · Hugging Face

Download file

transformer_2/5.safetensors

safetensors · 1.96 GB · SHA-256 e658e5e45271…7617 · Hugging Face

Download file

transformer_2/6.safetensors

safetensors · 1.96 GB · SHA-256 93cebfd35a90…054a · Hugging Face

Download file

transformer_2/7.safetensors

safetensors · 1.96 GB · SHA-256 3d73d1cbb1ac…73c7 · Hugging Face

Download file

transformer_2/8.safetensors

safetensors · 1.96 GB · SHA-256 3b1b5339b411…0db2 · Hugging Face

Download file

transformer_2/9.safetensors

safetensors · 1.96 GB · SHA-256 f8a23cc15756…28f6 · Hugging Face

Download file

transformer/0.safetensors

safetensors · MIXED (unverified) · 1.99 GB · SHA-256 077956a53e07…3fea · Hugging Face

Download file

transformer/1.safetensors

safetensors · BF16 (unverified) · 1.96 GB · SHA-256 03d004b2dae0…2ae6 · Hugging Face

Download file

transformer/10.safetensors

safetensors · 1.96 GB · SHA-256 9357ebc9ab1a…6b52 · Hugging Face

Download file

transformer/11.safetensors

safetensors · 1.96 GB · SHA-256 1323c2e22d6f…f8b3 · Hugging Face

Download file

transformer/12.safetensors

safetensors · 1.96 GB · SHA-256 07c337852a24…5bd8 · Hugging Face

Download file

transformer/13.safetensors

safetensors · 1.07 GB · SHA-256 72b0a3b4542e…9772 · Hugging Face

Download file

transformer/2.safetensors

safetensors · 1.96 GB · SHA-256 c3560d3937a7…e5ba · Hugging Face

Download file

transformer/3.safetensors

safetensors · 1.96 GB · SHA-256 10e2f5e5243f…32c6 · Hugging Face

Download file

transformer/4.safetensors

safetensors · 1.96 GB · SHA-256 d0517b2b5bdc…a5ae · Hugging Face

Download file

transformer/5.safetensors

safetensors · 1.96 GB · SHA-256 4abb152a6668…9de3 · Hugging Face

Download file

transformer/6.safetensors

safetensors · 1.96 GB · SHA-256 b36b0e4c7067…a717 · Hugging Face

Download file

transformer/7.safetensors

safetensors · 1.96 GB · SHA-256 730b11f1d1cf…b9b4 · Hugging Face

Download file

transformer/8.safetensors

safetensors · 1.96 GB · SHA-256 4d3ddbe5a34b…3158 · Hugging Face

Download file

transformer/9.safetensors

safetensors · 1.96 GB · SHA-256 a9b6614e8094…13ec · Hugging Face

Download file

vae/0.safetensors

safetensors · 242 MB · SHA-256 e3d43fd7e6e6…0411 · Hugging Face

Download file

Source excerpts

2 excerpts
ReadmeAug 24huggingface.co/AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16

--- license: apache-2.0 base_model: Wan-AI/Wan2.2-T2V-A14B-Diffusers pipeline_tag: text-to-video library_name: mlx-gen tags: - mlx - mlx-gen - mflux - apple-silicon - bf16 - wan - wan2.2 - video-generation - text-to-video - wan-a14b --- # wan2.2-t2v-a14b-diffusers-bf16 This repository contains BF16 MLX-Gen saved weights for [`Wan-AI/Wan2.2-T2V-A14B-Diffusers`](https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B-Diffusers). It is designed for local Apple Silicon inference with [`mlx-gen`](https://github.com/lpalbou/mlx-gen). It uses the mflux/MLX saved-weight layout. It is not a Diffusers or Transformers `from_pretrained()` checkpoint. ## Source Model Original model: [`Wan-AI/Wan2.2-T2V-A14B-Diffusers`](https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B-Diffusers). This prepared derivative follows the Apache 2.0 license of the source model. ## Precision This package stores the Wan A14B T2V transformer and VAE weights for MLX-Gen BF16 runtime use. The UMT5 text encoder, scheduler metadata, tokenizer files, and model index are included in the prepared folder. ## Validation Measured on 2026-06-04 with `mlx-gen 0.18.9` on Apple Silicon. The upstream Diffusers source snapshot measured about 118 GiB in the local Hugging Face cache before preparing these packages. The table below reports prepared-package generation from model init through MP4 save and post-save video-health validation. Validation profile: `384x224`, 33 frames, 12 denoising steps, guidance `4`, guidance-2 `3`, 8 fps, seed `4242`, `--low-ram`. | Package | Disk | Full-Process Physical Peak | Max RSS | MLX Peak | Total Time | Video Health | |---|---:|---:|---:|---:|---:|---| | This BF16 package | 64.3 GiB | 33.0 GiB | 31.8 GiB | 27.7 GiB | 152.7 s | 33/33 frames, 384x224, 8 fps, temporal delta 1.3 | | Mixed q8/BF16 package | 39.7 GiB | 20.7 GiB | 19.5 GiB | 15.5 GiB | 154.8 s | 33/33 frames, 384x224, 8 fps, temporal delta 1.4 | Physical peak is Darwin `ri_phys_footprint` sampled for the full process. The validation is intentionally small and repeatable; it is not a claim that every full-size `1280x720`, 81-frame, 40-step job has the same memory or timing profile. ## Usage ```bash python -m pip install -U mlx-gen mlxgen download --model AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16 mlxgen generate \ --model AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16 \ --task text-to-video \ --prompt "...

NoteAug 24huggingface.co/AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16

Source context: 0 downloads · 0 likes · Pipeline text-to-video · Library mlx-gen · Repo AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16