Check my GPU

LLMs ·

Nemotron 3.5 Lightning 30B-A3B

Parameters
32B (3.0B active)
Architecture
Mixture of experts
Max context
256K tokens
License
openmdw-1.1
Released
2026-08

How to run it

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

ollama run nemotron-3.5-lightning:30b

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

huggingface.co/unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF (opens in a new tab)

ollama run hf.co/unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF

Original weights · Hugging Face

Full-precision files from NVIDIA, for fine-tuning, vLLM, Transformers.

huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from NVIDIA. The weights themselves download from Hugging Face.

github.com/NVIDIA-NeMo/Nemotron (opens in a new tab)

git clone https://github.com/NVIDIA-NeMo/Nemotron

NVIDIA’s Nemotron repo with recipes for the whole family.

Links checked 5 days ago against Hugging Face, Ollama and GitHub.

Memory needed

Weights plus KV cache and runtime overhead. Leave a little extra free for the operating system and other apps.

KV cache: hybrid Mamba-Transformer: only the attention layers keep a KV cache.

Quantization Quality Weights Total at 8K context Total at 32K context Download
Q2_K Low 12.6 GB 13.3 GB 13.4 GB GGUF (IQ2_M) (opens in a new tab)
Q3_K_M Low 15.5 GB 16.1 GB 16.3 GB GGUF (UD-Q3_K_XL) (opens in a new tab)
Q4_K_M Good 19.3 GB 19.9 GB 20.1 GB GGUF (opens in a new tab)
Q5_K_M High 22.4 GB 23.1 GB 23.2 GB GGUF (opens in a new tab)
Q6_K High 25.9 GB 26.5 GB 26.7 GB GGUF (UD-Q6_K_XL) (opens in a new tab)
Q8_0 Very high 33.5 GB 34.1 GB 34.3 GB GGUF (opens in a new tab)
F16 Full 63.2 GB 63.8 GB 64.0 GB GGUF · folder (opens in a new tab) Original weights (opens in a new tab)

Best hardware for Nemotron 3.5 Lightning 30B-A3B

The least expensive GPU in our catalog that runs it without offloading is the AMD Radeon RX 7600 XT (launch price $329) at Q2_K.

Grade · points Hardware Quant Memory Speed Fit
87 ptsExcellent (S) · 87 of 100 points: Model quality 32.4/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15 Apple M4 Max (64 GB) Q8_0 38.0 GB / 48.0 GB 49.0 tok/s Runs well
87 ptsExcellent (S) · 87 of 100 points: Model quality 32.4/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15 NVIDIA RTX PRO 6000 Blackwell F16 68.6 GB / 96.0 GB 86.0 tok/s Runs well
86 ptsExcellent (S) · 86 of 100 points: Model quality 32.1/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 14.6/15 Apple M1 Max (64 GB) Q6_K 30.4 GB / 48.0 GB 48.9 tok/s Runs well
86 ptsExcellent (S) · 86 of 100 points: Model quality 31.4/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15 NVIDIA GeForce RTX 5090 Q5_K_M 24.7 GB / 32.0 GB 303 tok/s Runs well
86 ptsExcellent (S) · 86 of 100 points: Model quality 31.4/45, Memory fit 25.0/25, Speed 14.9/15, Memory use 15.0/15 Apple M4 Pro (48 GB) Q5_K_M 26.0 GB / 36.0 GB 39.7 tok/s Runs well
84 ptsGreat (A) · 84 of 100 points: Model quality 30.8/45, Memory fit 24.0/25, Speed 15.0/15, Memory use 15.0/15 AMD Radeon RX 7900 XTX Q4_K_M 21.1 GB / 24.0 GB 169 tok/s Tight fit
84 ptsGreat (A) · 84 of 100 points: Model quality 30.8/45, Memory fit 24.0/25, Speed 15.0/15, Memory use 15.0/15 NVIDIA GeForce RTX 3090 Q4_K_M 21.1 GB / 24.0 GB 191 tok/s Tight fit
84 ptsGreat (A) · 84 of 100 points: Model quality 30.8/45, Memory fit 24.0/25, Speed 15.0/15, Memory use 15.0/15 NVIDIA GeForce RTX 4090 Q4_K_M 21.1 GB / 24.0 GB 206 tok/s Tight fit
82 ptsGreat (A) · 82 of 100 points: Model quality 32.4/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 10.0/15 Apple M4 Max (128 GB) Q8_0 41.8 GB / 96.0 GB 49.0 tok/s Runs well
79 ptsGreat (A) · 79 of 100 points: Model quality 32.4/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 7.3/15 Apple M2 Ultra (192 GB) Q8_0 45.6 GB / 144 GB 71.7 tok/s Runs well

Check it on your own hardware

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.