Skip to content
localmodel.run

text model · Nemotron · macOS

Can I run Nemotron 3.5 Bolt E4B on Apple M5 Pro (48GB)?

Compatibility verdict VRAM threshold engine
Yes, it runs GPU accelerated ~33 tok/s est.

Yes. Nemotron 3.5 Bolt E4B runs on Apple M5 Pro (48GB) at Q4_K_M (~8.9 GB of ~32 GB usable).

Needs ~8.9 GB Device usable ~32 GB

Runs at Q4_K_M using ~8.9 GB of ~32 GB usable. You have room for FP16 for higher quality.

That figure is at a 4k context and moves about ±15% as context length changes. Apple M5 Pro (48GB) leaves ~23.1 GB of headroom, room to step up to FP16 for higher quality.

Q4_K_M needed
~8.9 GB
Usable on device
~32 GB
Device memory
48 GB
Best quant
Q4_K_M
Share on X Share on Reddit

Which quant fits

Quant ladder vs ~32 GB usable
Q2_K
~4.9 GB
Q3_K_M
~5.4 GB
Q4_K_M
~8.9 GB
Q5_K_M
~7.2 GB
Q6_K
~8.1 GB
Q8_0
~10.6 GB
FP16
~17.1 GB
The line marks Apple M5 Pro (48GB)'s ~32 GB budget; rungs past it are too large.

Run it

Install commands macOS

Loads the Q4_K_M weights.

Ollama
$ ollama run nemotron-3.5-bolt:e4b-q4_K_M
Model Nemotron
Parameters
8B
Q4_K_M size
7.37 GB
Q8_0 size
9.09 GB
Context
64k
Ollama tag
nemotron-3.5-bolt:e4b-q4_K_M
Full Nemotron 3.5 Bolt E4B requirements →
Device macOS
Memory
48 GB unified
Usable for weights
~32 GB
Best runtime
MLX direct / Ollama (MLX backend)
Best models for Apple M5 Pro (48GB) →

You could also run

Run Nemotron 3.5 Bolt E4B on other hardware

Nvidia GeForce RTX 3060 (12GB)Nvidia GeForce RTX 4070 (12GB)Nvidia GeForce RTX 5070 (12GB)Apple M2 (16GB)Apple M4 (16GB)Nvidia GeForce RTX 4060 Ti (16GB)

FAQ

Can Apple M5 Pro (48GB) run Nemotron 3.5 Bolt E4B?

Yes. Nemotron 3.5 Bolt E4B runs on Apple M5 Pro (48GB) at Q4_K_M (~8.9 GB of ~32 GB usable).

How much memory does Nemotron 3.5 Bolt E4B need?

Apple M5 Pro (48GB) has room to spare. At Q4_K_M the weights are ~7.37 GB; with KV cache and runtime overhead, budget ~8.9 GB at a 4k context.

What is the best tool to run Nemotron 3.5 Bolt E4B on macOS?

LM Studio for a simple setup; mlx-lm for the most speed. vLLM is NOT a Mac tool, it is a CUDA/Linux serving engine. Unified memory is not a fixed VRAM slice; ~70% is usable for weights.

Embed this

Nemotron 3.5 Bolt E4B on Apple M5 Pro (48GB) compatibility badge A live badge for your model card or README, updated as the data is.
Markdown
$ [![Nemotron 3.5 Bolt E4B on Apple M5 Pro (48GB)](https://localmodel.run/badge/nemotron-3.5-bolt-e4b/apple-m5-pro-48gb.svg)](https://localmodel.run/can-i-run/nemotron-3.5-bolt-e4b/apple-m5-pro-48gb)

Sources

Weights are measured from GGUF files; KV cache and overhead are computed, so totals can vary ~15% with context and runtime. Any tok/s is a bandwidth estimate. See methodology.