Jun Kim, oMLX creator and maintainer, joins Hugging Face to support the MLX community
Source
huggingface.co
Date
Context
Hugging Face is funding continued work on oMLX, software for running language models on Macs, by hiring its creator and maintainer. The company says the project will retain its Apache 2.0 license and existing leadership. The immediate change is funded maintenance; faster development is an aim.
One planned focus is making it easier to turn Transformers model definitions into reference implementations for MLX, the framework optimized for Apple Silicon. Different applications could then use that shared implementation and concentrate on their own features. This targets the work required to bring newly supported models into local applications.
Key takeaways · AI-distilled
Hugging Face names a concrete integration goal: streamlining the path from a transformers model definition to a reference MLX implementation that other engines, including mlx-lm, mlx-vlm, and LMStudio, can consume without each building its own port.
Hugging Face frames oMLX as a testbed that upstreams improvements into mlx-lm and mlx-vlm rather than replacing them, describing the goal as strengthening those libraries instead of displacing them.
Jun Kim keeps oMLX under its existing Apache 2.0 license and retains project leadership after the hire, which funds his prior volunteer role rather than installing new ownership.
The post names ongoing collaboration partners beyond MLX itself, saying Hugging Face wants to strengthen ties with mlx-lm, mlx-vlm, and LM Studio's teams, naming maintainers Cheng, Prince, and Yagil directly.
