🔍 Read the full analysis: How Jun Kim’s Move To Hugging Face Could Support The MLX Community on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Jun Kim, creator and maintainer of the MLX inference engine oMLX, has joined Hugging Face to work on Apple’s MLX ecosystem. Hugging Face says the project will receive funded maintenance while remaining Apache 2.0 licensed and under Kim’s leadership; timelines and specific technical deliverables have not been announced.
Jun Kim, creator and maintainer of the open-source oMLX inference engine, has joined Hugging Face to work on Apple’s MLX ecosystem, as detailed in the original analysis. Hugging Face said oMLX will move from a side project to a funded, maintained effort, while keeping its Apache 2.0 license and Kim’s leadership.
Hugging Face said the new arrangement is intended to give oMLX greater stability and faster development. The company said dedicated support will help Kim guide contributors and develop the project over the long term. It did not provide financial terms or describe the scope of the funding.
The company also named a technical priority: making it easier to turn a model definition from its Transformers library into a reference MLX implementation that multiple inference engines can use. Hugging Face said the aim is to make new Transformers models easier to run on MLX. It did not announce a delivery date or specific models for this work.
Hugging Face said it wants to work more closely with projects including mlx-lm, mlx-vlm and LM Studio, and with the teams behind them. It described oMLX as a place to explore ideas built on foundational MLX libraries, with potential work to be contributed upstream where appropriate. Those are stated intentions; the announcement did not specify contributions already planned or agreed.
More Support for Mac AI Tools
For developers and users running models on Apple Silicon, a funded maintainer could make oMLX a more consistently supported option for local inference. That matters because oMLX is maintained by Kim, and Hugging Face’s announcement places the work within a company-backed effort rather than leaving it solely as a side project.
The change may also benefit the wider MLX ecosystem if technical work is shared across projects. Hugging Face has said it wants to improve the path from Transformers model definitions to MLX implementations and collaborate with other teams. Whether that produces shared tools or upstream changes remains to be seen; the announcement establishes direction, not results.
How oMLX Fits Into MLX
MLX is Apple’s machine-learning framework designed for Apple Silicon. Released in late 2023, it supports running and training models locally on Apple hardware. Community projects such as mlx-lm and mlx-vlm build model support on top of the framework; oMLX is an inference engine in that ecosystem.
Hugging Face said it has supported MLX since the framework’s release and that its Hub is a place to discover and contribute MLX models. The company framed Kim’s hire as part of its investment in local, on-device AI. It also cited earlier collaboration with MLX-related projects, including mlx-lm, mlx-vlm and LM Studio.
“We are completely invested in local AI, and MLX is a central piece of the ecosystem.”
— Hugging Face
Details Still Awaiting Release
The announcement does not give Kim’s job title, team placement or start date, or specify the amount and duration of funding for oMLX. It also provides no schedule for the Transformers-to-MLX work, stability changes or faster releases.
Hugging Face’s plans to collaborate with other MLX projects and upstream work are described as intentions. It is not yet clear which changes will be delivered, when they will appear, or how the arrangement will affect oMLX’s roadmap over time.
Future oMLX releases and repository updates should show whether funded maintenance leads to a steadier development pace or new stability work. Another milestone to watch is whether Hugging Face publishes a practical path for turning Transformers model definitions into MLX implementations usable by different inference engines.
Further updates may clarify how Kim and Hugging Face will work with mlx-lm, mlx-vlm, LM Studio and other MLX teams, and whether any oMLX changes are contributed to shared libraries. The company said it will continue supporting MLX models on the Hub, but gave no timetable for these developments.
Key Questions
What did Jun Kim announce?
Hugging Face announced that Kim, the creator and maintainer of oMLX, has joined the company to work on Apple’s MLX ecosystem.
Will oMLX remain open source?
According to Hugging Face, oMLX will remain under the Apache 2.0 license, and Kim will continue leading the project.
What technical work is Hugging Face proposing?
The company wants to streamline the path from a Transformers model definition to a reference MLX implementation that multiple inference engines can use. It has not announced a timeline.
When will users see changes?
Hugging Face has not given a release schedule or a date for the planned technical work. Future oMLX releases and project updates will show how the new support affects development.
Primary source: Hugging Face · via ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
