Grux AI Download

How do I run NVIDIA Nemotron on my Mac?

Install Ollama and run ollama run nemotron-3-nano:4b, a 2.8GB to 2.9GB download of the smallest Nano, or ollama run nemotron-3-nano:30b, an 18GB to 24GB mixture-of-experts model with 3.5B active parameters. One trap: a bare ollama run nemotron-3.5-bolt pulls Bolt 74B-A9B, a 25GB download, and Ollama's page says Apple Silicon and Metal are not currently supported for that model. On a Mac, name the small one, ollama run nemotron-3.5-bolt:e4b, 7.2GB to 7.9GB and, Ollama says, supported on Apple Silicon. Nemotron 3.5 Lightning is a 30B model with 3B active parameters, 19GB to 25GB. No Nemotron is in Grux's catalogue, so this site rates none on any Mac and quotes Ollama's download sizes only.

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

The Nemotron models on Ollama

ModelCommandOllama listsInput
Nemotron 3 Nano 4Bollama run nemotron-3-nano:4b2.8GB to 2.9GB, 256K contextText
Nemotron 3 Nano 30B, the defaultollama run nemotron-3-nano18GB to 24GB, 1M contextText
Nemotron 3.5 Lightning 30B, the defaultollama run nemotron-3.5-lightning19GB to 25GB, 1M contextText
Nemotron 3.5 Bolt E4Bollama run nemotron-3.5-bolt:e4b7.2GB to 7.9GB, 128K context; supported on Apple Silicon, Ollama saysText and image
Nemotron 3.5 Bolt 74B-A9B, the bare defaultollama run nemotron-3.5-bolt25GB, 256K context; Ollama says Apple Silicon and Metal are not currently supportedText and image

Ollama also lists tags marked MLX: Nemotron 3 Nano 4B and 30B at 2.9GB and 18GB, and Lightning 30B at 19GB, each with a 256K context. Ollama's page for these models does not explain the MLX tags; MLX vs GGUF on a Mac covers what is documented.

What the models are

The larger Nemotron models

Ollama lists Nemotron 3 Super at 87GB, with a cloud tag beside it, and Nemotron 3 Ultra with only a cloud tag. A cloud tag runs on Ollama's servers, not locally.

What this site does not say

Grux's catalogue lists no Nemotron model, so the Mac picker does not rate one and this page writes no rating for it on any memory size. Download sizes are Ollama's, and are usually smaller than the memory a model needs while running. Nothing on this page was run on a Mac.

Where to check this

ClaimSource
Nano: 4B and 30B sizes, 3.5B active of 30B, licence, languages, MLX tagsOllama library, nemotron-3-nano
Lightning: 30B with 3B active, 19GB to 25GB, MLX tagOllama library, nemotron-3.5-lightning
Bolt: E4B and 74B-A9B, the default tag, no Apple Silicon for 74B-A9B, licenceOllama library, nemotron-3.5-bolt
Super at 87GB with a cloud tag; Ultra cloud onlyOllama library, nemotron-3-super and nemotron-3-ultra

Questions

How do I run Nemotron on a Mac?
Install Ollama and run ollama run nemotron-3-nano:4b for the small one, or ollama run nemotron-3.5-bolt:e4b for Bolt E4B, which Ollama says is supported on Apple Silicon.
Why does ollama run nemotron-3.5-bolt not work on my Mac?
The bare name pulls Bolt 74B-A9B, and Ollama's page says Apple Silicon and Metal are not currently supported for it. Use the :e4b tag.
How big is Nemotron 3 Nano?
Ollama lists the 4B at 2.8GB to 2.9GB and the 30B, the default, at 18GB to 24GB.
Is Nemotron free to use?
Nano is governed by the NVIDIA Open Model License Agreement and Bolt by the OpenMDW License Agreement 1.1; read each licence for your use. Ollama's Lightning page states none.
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.