Hugging Face has announced that Jun Kim, the creator and primary maintainer of oMLX, will be joining the company as a full-time team member. oMLX is an open-source framework that wraps Apple's MLX library to make it easier for developers to build and run machine learning models on Apple Silicon devices. The move represents a significant investment by Hugging Face in supporting the growing local AI development ecosystem.
The transition will allow oMLX to evolve from a side project maintained by Kim into a fully-funded and professionally maintained framework. Despite the change, oMLX will continue to operate under the Apache 2.0 open-source license, and Kim will retain leadership of the project. Hugging Face emphasized its commitment to supporting accessible, locally-run AI tools as cloud-based inference faces growing scrutiny over cost and privacy concerns.
A key focus for the collaboration is accelerating the workflow for converting Transformers models—the industry-standard model definition format—into optimized MLX implementations that can be consumed by different inference engines. Hugging Face also signaled plans to strengthen partnerships with related MLX ecosystem projects including mlx-lm, mlx-vlm, and LMStudio, positioning oMLX as a testbed for innovations that can be upstreamed across the broader ecosystem.
Key Points
Jun Kim joins Hugging Face as full-time member to support oMLX, an open-source wrapper for Apple's MLX framework
oMLX transitions from side project to fully-funded initiative while remaining Apache 2.0 open source with Kim retaining leadership
Focus on streamlining conversion of Transformers models to optimized MLX implementations for consumption across inference engines
Partnership aims to strengthen ecosystem collaboration with related projects including mlx-lm, mlx-vlm, and LMStudio