NVIDIA

Nemotron 3 33B

NVIDIA·33B·Mistral·128K ctx·2026-08

Model sizeHardware requirementActual fit

Nemotron 3 33B needs ~29.2–54.7 GB of VRAM depending on quantization. The bar is what it needs; the marker is the card's VRAM. Speed figures below are planning estimates.

8 GB

RTX 4060 · 3060 Ti

29.2
8
Q4_K_M · Est. ~2 tok/sSlow · CPU offload

12 GB

RTX 4070 · 3060 12G

29.2
12
Q4_K_M · Est. ~4 tok/sSlow · CPU offload

16 GB

RTX 4080 · M-series 16G

29.2
16
Q4_K_M · Est. ~6 tok/sSlow · CPU offload

24 GB

RTX 4090 · 3090

29.2
24
Q4_K_M · Est. ~9 tok/sSlow · CPU offload

48 GB

A6000 · 2×24G

34.3
48
Q5_K_M · Est. ~46 tok/sComfortable

02  — Quantizations

QuantFile sizeEst. VRAM
Q4_K_M28 GB29.2 GB
Q5_K_M33.1 GB34.3 GB
Q8_053.5 GB54.7 GB

03  — Runtimes & use

llama.cpp · Ollama · LM Studio

General · Max intelligence

Memory and speed figures are planning estimates, not measured benchmarks. Memory includes a fixed allowance for KV cache and runtime overhead. Long-context prompts push VRAM use above the numbers shown here.