Models

NVIDIA Nemotron 3.5 Lightning 30B A3B BF16

NVIDIA · August 2026

Get

Hugging Face

A 33B, 0.7B active chat model from NVIDIA.

Rating: 2.7
Downloads: 2.4K
Likes: 1
Parameters: 33B, 0.7B active
Context: 1M
Licence: other

Too big for this Mac

Its smallest size needs 23 GB of memory, more than M4 · 16 GB can give it.

Ratings

2.7 out of 5, across the tests below.

KnowledgeMMLU-Pro: 81.9%
ScienceGPQA Diamond: 75.4%
Expert questionsHumanity's Last Exam: 11.7%
Software fixesSWE-bench Verified: 51.6%

MMLU-Pro: graduate-level questions in 14 subjects. GPQA Diamond: questions PhD scientists write for each other. Humanity's Last Exam: questions written to be beyond today's models. SWE-bench Verified: real GitHub issues fixed so the project's tests pass. Scores are the ones filed on the model's Hugging Face page or the board's own table, read 2026-11-05; some come from the maker.

NVIDIA Nemotron 3.5 Lightning 30B A3B BF16 runs well from a Mac with an M1 Pro and 32 GB of memory. 40 of 59 Mac configurations run it well.

Compare with

The nearest models in size.

Sizes

From mradermacher/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

Size	Download	Memory at 8K	Get
Q2_K	19 GB	19 GBestimated	Open
Q3_K_S	19 GB	19 GBestimated	Open
IQ4_XS	19 GB	19 GBestimated	Open
Q3_K_M	21 GB	21 GBestimated	Open
Q3_K_L	22 GB	22 GBestimated	Open
Q4_K_S	23 GB	23 GBestimated	Open
Q5_K_S	25 GB	25 GBestimated	Open
Q4_K_M	25 GB	26 GBestimated	Open
Q5_K_M	27 GB	27 GBestimated	Open
Q6_K	35 GB	35 GBestimated	Open
Q8_0	35 GB	35 GBestimated	Open

Run It Anywhere

NVIDIA Nemotron 3.5 Lightning 30B A3B BF16 is an open model, so any app that runs GGUF files can run it. These take the Q2_K size from mradermacher straight from Hugging Face.

LM StudioMac, Windows, LinuxOpen
JanMac, Windows, LinuxOpen
MstyMac, Windows, LinuxOpen
OllamaMac, Windows, Linuxollama run hf.co/mradermacher/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16-GGUF:Q2_K
llama.cppMac, Windows, Linuxllama-server -hf mradermacher/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16-GGUF:Q2_K
Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/mradermacher/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16-GGUF:Q2_K

On a PC Graphics Card

Video memory	Largest size that fits	Download	Memory at 8K
6 GB	None worth running		needs 19 GB
8 GB	None worth running		needs 19 GB
10 GB	None worth running		needs 19 GB
11 GB	None worth running		needs 19 GB
12 GB	None worth running		needs 19 GB
16 GB	None worth running		needs 19 GB
20 GB	IQ4_XS	19 GB	19 GB
24 GB	Q4_K_S	23 GB	23 GB
32 GB	Q5_K_M	27 GB	27 GB
48 GB	Q8_0	35 GB	35 GB

The largest size, of those that keep three bits a weight or more, whose memory at an 8K chat is under 92% of the card's; the rest is the desktop's. Estimated from the files, the same way as for a Mac; speed on a PC is not estimated yet. A size that doesn't fit the card still runs, slower, with part of it in ordinary memory.

On every Mac

Mac	Size	Fit	Memory	Writes
M6 · 16 GB	Q2_K	Too big	23 GB
M6 · 24 GB	Q2_K	Too big	23 GB
M6 · 32 GB	Q3_K_M	Runs well	21 GB	63
M5 · 16 GB	Q2_K	Too big	23 GB
M5 · 24 GB	Q2_K	Too big	23 GB
M5 · 32 GB	Q3_K_M	Runs well	21 GB	67
M5 Pro · 24 GB	Q2_K	Too big	23 GB
M5 Pro · 48 GB	Q4_K_S	Runs great	23 GB	99
M5 Pro · 64 GB	Q5_K_M	Runs great	29 GB	92
M5 Max · 36 GB	Q4_K_S	Runs well	23 GB	121
M5 Max · 48 GB	Q4_K_S	Runs great	23 GB	136
M5 Max · 64 GB	Q5_K_M	Runs great	29 GB	130
M5 Max · 128 GB	Q8_0	Runs great	39 GB	139
M5 Ultra · 96 GB	Q8_0	Runs great	39 GB	178
M5 Ultra · 256 GB	Q8_0	Runs great	39 GB	178
M5 Ultra · 512 GB	Q8_0	Runs great	39 GB	178
M4 · 16 GB	Q2_K	Too big	23 GB
M4 · 24 GB	Q2_K	Too big	23 GB
M4 · 32 GB	Q3_K_M	Runs well	21 GB	32
M4 Pro · 24 GB	Q2_K	Too big	23 GB
M4 Pro · 48 GB	Q4_K_S	Runs great	23 GB	42
M4 Pro · 64 GB	Q5_K_M	Runs great	29 GB	40
M4 Max · 36 GB	Q4_K_S	Runs well	23 GB	45
M4 Max · 48 GB	Q4_K_S	Runs great	23 GB	48
M4 Max · 64 GB	Q5_K_M	Runs great	29 GB	47
M4 Max · 96 GB	Q8_0	Runs great	39 GB	49
M4 Max · 128 GB	Q8_0	Runs great	39 GB	52
M3 · 8 GB	Q2_K	Too big	23 GB
M3 · 16 GB	Q2_K	Too big	23 GB
M3 · 24 GB	Q2_K	Too big	23 GB
M3 Pro · 18 GB	Q2_K	Too big	23 GB
M3 Pro · 36 GB	Q4_K_S	Runs well	23 GB	38
M3 Max · 36 GB	Q4_K_S	Runs well	23 GB	48
M3 Max · 48 GB	Q4_K_S	Runs great	23 GB	52
M3 Max · 64 GB	Q5_K_M	Runs great	29 GB	51
M3 Max · 96 GB	Q8_0	Runs great	39 GB	50
M3 Max · 128 GB	Q8_0	Runs great	39 GB	55
M3 Ultra · 96 GB	Q8_0	Runs great	39 GB	63
M3 Ultra · 256 GB	Q8_0	Runs great	39 GB	63
M3 Ultra · 512 GB	Q8_0	Runs great	39 GB	63
M2 · 8 GB	Q2_K	Too big	23 GB
M2 · 16 GB	Q2_K	Too big	23 GB
M2 · 24 GB	Q2_K	Too big	23 GB
M2 Pro · 16 GB	Q2_K	Too big	23 GB
M2 Pro · 32 GB	Q3_K_M	Runs well	21 GB	43
M2 Max · 32 GB	Q3_K_M	Runs well	21 GB	52
M2 Max · 64 GB	Q5_K_M	Runs great	29 GB	50
M2 Max · 96 GB	Q8_0	Runs great	39 GB	54
M2 Ultra · 64 GB	Q5_K_M	Runs great	29 GB	56
M2 Ultra · 128 GB	Q8_0	Runs great	39 GB	62
M2 Ultra · 192 GB	Q8_0	Runs great	39 GB	62
M1 · 8 GB	Q2_K	Too big	23 GB
M1 · 16 GB	Q2_K	Too big	23 GB
M1 Pro · 16 GB	Q2_K	Too big	23 GB
M1 Pro · 32 GB	Q3_K_M	Runs well	21 GB	30
M1 Max · 32 GB	Q3_K_M	Runs well	21 GB	34
M1 Max · 64 GB	Q5_K_M	Runs great	29 GB	33
M1 Ultra · 64 GB	Q5_K_M	Runs great	29 GB	36
M1 Ultra · 128 GB	Q8_0	Runs great	39 GB	40

Writes is words a second, estimated from the Mac's memory speed and the model's size; how, and how wrong.

Information

Maker: NVIDIA
Released: August 2026
Parameters: 33B, 0.7B active
Longest chat: 1M tokens
Reads pictures: No
Licence: other
Downloads this month: 2,402
Likes: 1
Sizes: 11, 19 to 35 GB
Kind: Maker's release
Source: nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16

Badge for the model card

Shows on NVIDIA Nemotron 3.5 Lightning 30B A3B BF16's Hugging Face card which Macs run it well, linked to this page. It updates every morning.

Files, downloads and likes read from Hugging Face on November 5, 2026 and refreshed every morning. Memory and speed are estimates until someone running Outmake measures this model on that Mac.