Running a local MLX music model via Codex pushes M4 MacBook Air to its limits

Dimillian · x · 2026-09-08

Developer Dimillian is using Codex to drive a local MLX model on an M4 MacBook Air to generate music for his app Evergrow. The M4 is visibly struggling, with the ChatGPT app's heavy RAM usage attributable to the local model workload.

Original post →

More from Infra

Infra channel →