Grux AI Download

Which local LLM can a 96 GB Mac run?

The Grux AI app's fit rule rates every model in its catalogue for a 96 GB Mac below, by the memory the model needs while it runs against about three quarters of the Mac's unified memory. The model the app starts a 96 GB Mac on is marked, with its Ollama pull line. The summary below the table is plain text to copy, and this address is the 96 GB result of the Mac picker, so it can be linked or shared. No speed is stated: speed depends on the chip, and only one Mac has been measured. A rating here says whether a model fits in memory, and nothing about how fast it runs or how well it answers.

Download Grux AI 3.3.1, free, for macOS 14 or later on Apple silicon.

Grux AI 3.3.1 · sources read 2026-11-19 · published 2026-11-19

What a 96 GB Mac can run

96 GB Mac: about 72 GB is available to a model.

The app would start this Mac on Grux Local 9B (great fit): ollama pull gruxai.com/grux/grux-local:9b

ModelMemory it needsFit
Grux Local 9BOllama 0.40.8 or newer7.25 GBGreat fit
Grux Local 4BOllama 0.40.8 or newer4.44 GBGreat fit
Gemma 4 26BOllama 0.30.9 or newer19.2 GBGreat fit
Grux Local 2BOllama 0.40.8 or newer2.65 GBGreat fit
Gemma 4 12BOllama 0.30.9 or newer9.4 GBGreat fit
Qwen 3.8 27BOllama 0.32.12 or newer17.4 GBGreat fit
Qwen 3.5 4BOllama 0.30.0 or newer4.4 GBGreat fit
Qwen 3.5 9BOllama 0.30.0 or newer6.2 GBGreat fit
Qwen 3.5 35BOllama 0.30.0 or newer35 GBGreat fit
Qwen 3 Coder 30B28 GBGreat fit
GPT-OSS 20BOllama 0.30.0 or newer20 GBGreat fit
Mistral Small 3.2 24B22 GBGreat fit
Qwen 3.5 2BOllama 0.30.0 or newer2.6 GBGreat fit
Llama 3.2 3B3 GBGreat fit
Qwen 3 VL 4BOllama 0.30.0 or newer6 GBGreat fit
Llama 3.3 70B60 GBTight

Too big for this Mac: GPT-OSS 120B (88 GB).

A summary to copy

Select the box and copy it. It is the result above as plain text, and every rating in it is the Mac picker's.

Which local LLMs a 96 GB Mac can run, by the Grux AI app's fit rule (about 72 GB available to a model)
Start with: Grux Local 9B (ollama pull gruxai.com/grux/grux-local:9b)
Great fit: Grux Local 9B, Grux Local 4B, Gemma 4 26B, Grux Local 2B, Gemma 4 12B, Qwen 3.8 27B, Qwen 3.5 4B, Qwen 3.5 9B, Qwen 3.5 35B, Qwen 3 Coder 30B, GPT-OSS 20B, Mistral Small 3.2 24B, Qwen 3.5 2B, Llama 3.2 3B, Qwen 3 VL 4B
Good fit: none
Tight: Llama 3.3 70B
Too big: GPT-OSS 120B
Rated by memory only, not speed.
https://gruxai.com/answers/local-llm-for-a-96gb-mac/

What the next size up changes

From this size to the next one in the picker. The memory step chooser compares every larger size.

Mac memoryThe app starts it onWhat changes from 96 GB
128 GB32 GB moreGrux Local 9Bsame as on a 96 GB MacLlama 3.3 70B: Tight to Good fit; GPT-OSS 120B: Too big to Tight

Other memory sizes

8 GB, 16 GB, 18 GB, 24 GB, 32 GB, 36 GB, 48 GB, 64 GB, 128 GB

What this page does not cover

Where to check this

ClaimSource
The fit rule and every model's serving memoryGrux Cookbook.swift at the pinned commit
A Mac's model budget is a share of its unified memory, which Metal reports for each GPUApple, recommendedMaxWorkingSetSize

Questions

How does this page rate a model for a 96 GB Mac?
By the memory the model needs while it runs, against about three quarters of the Mac's 96 GB of unified memory. It is the Mac picker's rule and the Grux AI app's, and the table gives each model's rating.
Does this page say how fast a model runs?
No. It rates fit by memory. The one Mac measured for this site is a 16 GB Apple M4 Mac mini, on the Grux Local speed page.
Where do the ratings come from?
From the Grux AI app's own catalogue and fit rule, read from its public source at one pinned commit, shown in the sources table on this page.
Download Grux AI 3.3.1 Read the source

Free, MIT licensed. macOS 14 or later, Apple silicon. 29.1 MB, signed and notarized by Apple. No account, no server, no subscription.