cd /news/large-language-models/local-llm-on-macbook-m5-pro-totally-… · home topics large-language-models article
[ARTICLE · art-45408] src=discuss.huggingface.co ↗ pub= topic=large-language-models verified=true sentiment=· neutral

Local LLM on MacBook M5 Pro - Totally New to This!

A non-programmer named Tim is setting up a local LLM on a MacBook M5 Max with 128GB unified memory, using Docker Desktop with Model Runner, Open WebUI, and models like Gemma 4 and Qwen3 30B-A3B-Q4_k_m for daily use and deep research. He has built RAG knowledge collections for personal topics and seeks guidance on advancing his setup beyond Claude Pro.

read2 min views1 publishedJun 30, 2026

Ok to start with I am not a programmer or an IT person whatsoever. I have spent the last few months learning as much as I can about AI. I have been using Coursera to study topics on the various LLMS, CISSP, Python etc and continue to do so. I have a Claude Pro account that I am using to help me set this LLM up and have a temporary Gemini pro account via Coursera. I am trying to get a good general background on AI and security but I am still ignorant in so many ways.

Hardware:

Macbook M5 Max, 18 core cpu, 40 core gpu, 128GB unified memory, 4TB hd, OS Tahoe

Will eventually use Tailscale for remote access

Stack:

Docker Desktop using Docker Model Runner for local inference (full Metal GPU/unified memory access currently). Open webUI as chat interface, in Docker via Compose

Models:

Gemma 4 (~12B) - daily use

Qwen3 30B-A3B-Q4_k_m - deep research

RAG:

SentenceTransformers embedding, default function calling mode. Multiple topic-based knowledge collections (RV/camping, photography gear, truck, home equipment, etc) each one of these contain AI written md files with full manufacturer manuals pdf.

Other tools:

DrawThings - image/video generation

MacWhisper Pro - audio/video transcription

Kokoro TTS - local voice output

What is my objective?

So far this whole setup has been guided by Claude and has been a painful process but a learning one. I most likely went about this the wrong way and i’m ok with that but I feel like my next step is to make sure my models work the way I need them to and start using them more than Claude. I’m just not sure at my skill level where to go and what to do without it being too technical. I feel like there is a normal path that people follow but I don’t know what it is.

I will likely be asking lots of questions of you guys, thank you,

Tim

── more in #large-language-models 4 stories · sorted by recency
── more on @macbook m5 max 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/local-llm-on-macbook…] indexed:0 read:2min 2026-06-30 ·