Is 64GB RAM necessary on a new Mac when you already have a powerful local LLM desktop?
gappyvalley · reddit · 2026-08-30
A user with a powerful desktop LLM server (Dual RTX 50-series GPUs, 31GB VRAM) is debating whether to buy the 64GB version of the MacBook Pro M5.
Context:
- Desktop: Runs Qwen 3.8 27B efficiently; accessed remotely via Tailscale with great performance.
- Mac: Current M1 Pro 16GB is limiting, used mainly for Docker/K8s development and Moonlight streaming.
The Dilemma: Is there a meaningful benefit to running LLMs directly on the Mac versus connecting to the existing desktop? Does the jump to 64GB (enabling local Qwen 27B q8) justify the extra cost and one-month wait over the 48GB model? Community consensus leans towards 48GB being sufficient for development tasks.
More from Infra
- Full-stack AI EDA to disrupt chip design economics — ai · 2026-08-30
- NVIDIA's blockbuster earnings keep coming, but the stock has been stuck since April — thedealdirector · 2026-08-30
- Apple's Mac Biz Up 29%; OpenAI/Anthropic Buy Minis for AI Compute — Hesamation · 2026-08-30
- TrustScoreAgent: Open reputation registry for agent API calls — TrustScoreAgent · 2026-08-30
- LongCat-Flash-Lite-Sparse and Qwen Uncensored Models Released in GGUF — LLMFan46 · 2026-08-30
- TensorSharp integrates MiniMax H3 for local image-to-video inference — fuzhongkai · 2026-08-30