Instructions to use fukujusou/Anima-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use fukujusou/Anima-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir Anima-mlx fukujusou/Anima-mlx
- Diffusion Single File
How to use fukujusou/Anima-mlx with Diffusion Single File:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
Anima MLX Weights
This repository contains MLX-format converted weights for circlestone-labs/Anima.
The files are provided for non-commercial use only under the CircleStone Labs Non-Commercial License. This conversion is not an official product of CircleStone Labs LLC and has not been endorsed, approved, or validated by CircleStone Labs LLC.
Files
anima_mlx/
generate.py
tokenizers/
qwen25_tokenizer/
t5_tokenizer/
split_files/
diffusion_models/
anima-base-v1.0-mlx.safetensors
text_encoders/
qwen_3_06b_base-mlx.safetensors
vae/
qwen_image_vae-mlx.safetensors
No ComfyUI workflow JSON files or example images are included in this repository.
The included anima_mlx package is a minimal MLX inference runtime for these converted weights. The included tokenizer files reproduce the tokenizer paths expected by the runtime; ComfyUI itself is not required.
Source
- Base model:
circlestone-labs/Anima - Original license: CircleStone Labs Non-Commercial License
- Original model card notes that Anima is a derivative of NVIDIA Cosmos-Predict2-2B-Text2Image; users should also review the upstream NVIDIA OMLA terms referenced by the original model card.
Install
python -m pip install -r requirements.txt
This runtime requires Apple Silicon with MLX support.
Local Use
python -m anima_mlx.generate \
--prompt "test" \
--steps 1 \
--height 512 \
--width 512 \
--output outputs/anima-mlx-smoke.png
The default paths are:
--text-encoder split_files/text_encoders/qwen_3_06b_base-mlx.safetensors
--diffusion-model split_files/diffusion_models/anima-base-v1.0-mlx.safetensors
--vae split_files/vae/qwen_image_vae-mlx.safetensors
--comfy-root .
License
The CircleStone Model is licensed by CircleStone Labs LLC under the CircleStone Non-Commercial License. Copyright CircleStone Labs LLC.
IN NO EVENT SHALL CIRCLESTONE LABS LLC BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.
This repository distributes a modified format of the applicable CircleStone Model weights. Any rights to use the model or derivatives are granted directly by CircleStone Labs LLC pursuant to the included license.
Quantized
Model tree for fukujusou/Anima-mlx
Base model
nvidia/Cosmos-Predict2-2B-Text2Image
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js