oMLX Creator Jun Kim Joins Hugging Face to Advance MLX Ecosystem
Jun Kim, the creator and maintainer of oMLX, has joined Hugging Face effective September 2026. The move aims to stabilize oMLX development, accelerate contributions to Apple's…
- Jun Kim's full-time role at Hugging Face provides dedicated resources for oMLX, transitioning it from a side project to a funded open-source effort under Apache 2.0.
- Focus will include faster transformers-to-MLX conversions and upstreaming improvements to mlx-lm, mlx-vlm and related inference tools.
- Hugging Face positions the hire as strengthening the local AI ecosystem on Apple hardware while maintaining collaboration with projects like LM Studio.
Timeline of the Announcement
Apple introduced MLX as an open framework for machine learning on Apple Silicon in late 2023. Hugging Face quickly became the primary hub for discovering and sharing MLX-compatible models.
Jun Kim created and maintained oMLX as a side project, building on top of core MLX libraries to extend functionality for the community.
On September 22 2026 Hugging Face publicly announced that Jun Kim had joined the company to dedicate his efforts to supporting the broader MLX ecosystem through oMLX.
Impact on oMLX Development
Previously a part-time endeavor, oMLX will now benefit from full-time maintenance and organizational support. This change is expected to bring greater stability, faster iteration, and clearer long-term direction under Kim's continued leadership.
The license remains Apache 2.0 with no changes to governance. Contributors can expect the same open collaboration model while gaining resources to handle growing demand.
Broader Goals for MLX and Local AI
Hugging Face states its objective is to remove barriers for users running local AI models in any form. oMLX is viewed as a testbed for experimental features that can later be upstreamed to foundational libraries such as mlx-lm or mlx-vlm.
A concrete priority is streamlining the conversion process from Hugging Face transformers model definitions into efficient MLX implementations. This would allow different inference engines to focus on their specialized features while sharing a common reference implementation.
The company intends to deepen existing collaborations with maintainers of mlx-lm, mlx-vlm, LM Studio and other projects to better serve the local AI community on Apple hardware.
Context for Uncensored and Open Model Users
This development reinforces the importance of open, locally runnable tools that give users control over model inference without corporate gatekeeping. For those exploring expressive freedom in AI companions or avoiding model refusals on sensitive topics, robust MLX support on personal Apple Silicon hardware offers a practical path to greater independence.
In that spirit we recommend considering Muah AI when seeking an uncensored multimodal platform that combines chat, photos, voice, phone calls and advertised video capabilities in one hosted experience. Its positioning since the 2023 launch aligns with user choice in the uncensored AI space while complementing the open model ecosystem advanced by projects like MLX.
