What happened
Hugging Face announced that Jun Kim, creator and maintainer of oMLX, is joining its team, allowing oMLX to move from a side job to a fully maintained and funded project while staying Apache 2.0.
Why it matters
This is expected to bring stability and possibly faster development to oMLX, letting Kim better guide contributors and build for the long term.
What to watch
Hugging Face says it wants oMLX to serve as a testbed for new ideas and to streamline turning transformers model definitions into reference MLX implementations that different engines can use.
WHO IT HITSDevelopers and teams running local AI models on Apple Silicon, especially those using or contributing to oMLX, gain a more stable, funded project, while MLX engine and library maintainers may see a faster path from model definitions to MLX implementations.
Summaries like this, in your inbox every morning.
Hugging Face frames its decision as a continuation of support it has shown for MLX since the framework arrived as a Christmas present from Awni and Angelos in 2023. It calls itself the Hub where people find MLX models and contribute their own, and it describes open, local AI usage as accelerating.
The arrangement is unusual in that oMLX keeps its Apache 2.0 license and its existing lead, even as it gains funding and maintenance. Hugging Face expects oMLX to try out new ideas while building on dependencies such as mlx-lm and mlx-vlm, and it says it would like to upstream work wherever that makes sense. It also notes existing collaboration with projects including mlx-lm, mlx-vlm, and LMStudio, and hopes to strengthen ties with Cheng, Prince, Yagil, and their teams.
What this ultimately changes for the wider MLX ecosystem may hinge on how quickly the planned pipeline from transformers model definitions to reference MLX implementations takes shape, and on whether the strengthened relationships with other engine and library teams translate into shared work.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, LINE, or Slack.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
AWS launched Strands Harness, an open-source AI agent that runs locally or in any cloud, with downloads on Git…
MindWalk announced results from a production deployment of the open-source OpenFold3 structure prediction mode…
A LessWrong essay argues that when Anthropic and OpenAI publish safety or alignment research, it is often enti…
The Base Browser project, hosted at codeberg.org/basebrowserproject, released a hard fork of Firefox that it l…
Alibaba's Qwen AI team released Qwen-Image-2.1, an open-weight model with a 7-billion-parameter visual compone…
Tencent's Hunyuan Speech team and university researchers introduced Gander, which splits real-time conversatio…
