🔍 Read the full analysis: What Jun Kim’s Hugging Face Role Means For MLX Users on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- 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. The company says oMLX will receive funding and dedicated maintenance while remaining Apache 2.0 licensed 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, the company announced. Hugging Face says oMLX will move from a side project to a funded, fully 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 support faster development. The company said dedicated funding will allow Kim to guide contributors and work on the project for the long term. It did not disclose funding terms or say how many people, beyond Kim, will work on oMLX.
The announcement also describes a technical goal: make it easier to turn a model definition in Hugging Face Transformers into a reference implementation for MLX that multiple inference engines can use. Hugging Face said this could make it easier to run newer Transformers models on MLX. It gave no schedule or named models for this work.
Hugging Face said oMLX will remain Apache 2.0 licensed and that Kim will continue leading the project. The company also expressed interest in upstreaming work where appropriate and working with other MLX ecosystem projects, including mlx-lm, mlx-vlm and LM Studio. Those collaborations were presented as intentions; the post did not identify specific contributions already agreed or delivered.
More Support for Mac-Based Inference
For developers running models on Apple Silicon, a funded maintainer could make oMLX more dependable and sustain development beyond the capacity of a side project. That matters to users who want to run inference locally, where workloads can be handled on their own hardware rather than sent to a cloud service.
The announcement also links oMLX to a broader effort to connect Transformers model definitions with MLX implementations. If that work produces reusable paths across inference engines, it could reduce the effort required to make models available in Apple’s ecosystem. For now, that is a stated goal, not a demonstrated result.
Kim’s continued leadership and the unchanged license provide continuity for current users and contributors. Hugging Face’s institutional support may reduce the risk that development stalls because a single maintainer has limited time. Whether the project’s priorities continue to serve the wider MLX community will depend on the work and decisions that follow.
How oMLX Fits Into MLX
MLX is Apple’s machine learning framework designed for Apple Silicon. Hugging Face described it as a central part of its local AI efforts and said it has supported the framework since its release in late 2023. Projects such as mlx-lm and mlx-vlm provide model support and tools in the MLX ecosystem.
oMLX is an open-source inference engine created and maintained by Kim. Hugging Face’s Hub provides a place to discover and publish MLX models. The company framed Kim’s hire as part of its continuing investment in local, on-device AI and said it hopes to strengthen ties with teams across the MLX ecosystem.
““We are completely invested in local AI, and MLX is a central piece of the ecosystem.””
— Hugging Face
Funding, Timing and Deliverables
The announcement did not specify Kim’s job title, team placement or start date, and it did not disclose the amount or scope of funding for oMLX. It also gave no timeline for the Transformers-to-MLX work or the expected pace of releases.
Hugging Face’s interest in upstreaming improvements and collaborating with other project teams remains a stated intention. The company did not name committed upstream changes or describe agreements with those teams. It is not yet clear how quickly users will see changes in oMLX or whether the planned work will benefit other MLX tools.
Users can follow oMLX releases and repository updates for evidence of ongoing maintenance, stability improvements and a change in development pace. Progress toward a reusable path from Transformers definitions to MLX implementations will be a key sign of whether the technical plan is taking shape.
Further updates may clarify how Hugging Face will work with mlx-lm, mlx-vlm, LM Studio and other MLX contributors, and whether oMLX improvements are shared with projects it builds on. The announcement provided no dates for those developments.
Key Questions
Who is Jun Kim?
Jun Kim created and maintained oMLX, an open-source inference engine built for Apple’s MLX ecosystem. Hugging Face said he has joined the company to work on that ecosystem.
Will oMLX’s license change?
Hugging Face said oMLX will remain under the Apache 2.0 license.
What technical work does Hugging Face plan?
The company described a goal of making it easier to turn Transformers model definitions into reference MLX implementations that different inference engines can use. It did not provide a timeline or list specific models.
When will users see changes?
The announcement did not give a schedule. Users can watch oMLX release notes and repository updates for signs of new maintenance work and technical changes.
Primary source: Hugging Face · via ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
