The short version
An Apple Silicon local-model runtime with UMA and NVMe caching optimizations.
From the HackerLinks archive
oMLX
An Apple Silicon local-model runtime with UMA and NVMe caching optimizations.
At a glance:
First seen:2026-06-12
Last seen:2026-08-11
Times seen:2
Website:omlx.ai
Why it caught our attention
It was explicitly recommended as a high-performance Apple Silicon project for local model work.
Where it surfaced on Hacker News
2026-08-11
“try oMLX or vMLX - both great projects that offer some amazing performance optimizations for Apple Silicon”
reilly3000 · recommendation
Direct HN comment
Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
2026-06-12
Editorial paraphrase
The thread praised its model selection, fast downloads, and launching Claude Code and Codex on Apple Silicon.
Original threadHow to setup a local coding agent on macOS
Also surfaced in this discussion
Muse Glimmer 30B
Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
2026-08-11
vMLX
Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
2026-08-11