Ask HN: MacBook vs. Dedicated GPU for LLM

A question that keeps coming up on Hacker News: can you run large language models on a MacBook, and how does it compare to a dedicated GPU? The short answer reflects a real developer reality. Apple's M-series chips have enough horsepower to run smaller LLMs locally—think seven to thirteen billion parameters—but they hit hard limits on memory and thermal performance with anything bigger. Dedicated graphics processors offer raw speed and can handle much larger models, but they require server hardware or a workstation. For individual developers doing local experiments and prototyping, a MacBook works fine. For production workloads or running bigger models at scale, you'll want dedicated GPU hardware. It's the classic trade-off: portability and simplicity versus raw performance.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton