FitMyLLM
NVIDIA/Dense

NVIDIANemotron 3 Nano 4B

Nemotron 3 Nano — compact model with tool use and function calling.

chatcodingreasoningmathtool_useThinkingTool Use
3.97B
Parameters
256K
Context length
20
Benchmarks
6
Quantizations
100K
HF downloads
Architecture
Dense
Released
2026-03-16
Layers
42
KV Heads
8
Head Dim
128
Family
nemotron

Quantization Options

Context length:
QuantBitsVRAM @ 16KQuality
Q4_K_M4.89
3.1 GB
2.9 + 0.2 KV
good
Q5_K_S5.57
3.4 GB
3.3 + 0.2 KV
good
Q5_K_M5.7
3.5 GB
3.3 + 0.2 KV
good
Q6_K6.56
3.9 GB
3.7 + 0.2 KV
excellent
Q8_08.5
4.9 GB
4.7 + 0.2 KV
lossless
FP1616
8.6 GB
8.4 + 0.2 KV
lossless

Select your GPU above to see speed estimates and compatibility for each quantization.

Deploying for a team or in production? Size GPUs, cost & scaling in Enterprise →
Ready to run this?Rent by the hour

Rent a GPU and run Nemotron 3 Nano 4B now

Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.

Community Ratings

Loading ratings...

Benchmarks (20)

Run this model

Easiest way to get started·Beginners
Docs ↗
curl -fsSL https://ollama.com/install.sh | sh
$ollama run nemotron-3-nano:4b

Ollama's default build of this model. For Q4_K_M specifically, pick a tag from ollama.com/library/nemotron-3-nano/tags.

Setup guide
>_

Auto-setup with fitmyllm CLI

Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.

pip install fitmyllmthen run fitmyllmLearn more
Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines

GPUs that can run this model

At Q4_K_M quantization. Sorted by minimum VRAM.

NVIDIA Tesla C2050
3 GB VRAM • 144 GB/s
NVIDIA
NVIDIA Tesla M2050
3 GB VRAM • 148 GB/s
NVIDIA
NVIDIA Tesla S2050
3 GB VRAM • 148 GB/s
NVIDIA

Find the best GPU for Nemotron 3 Nano 4B

Build Hardware for Nemotron 3 Nano 4B

Nemotron 3 Nano — compact model with tool use and function calling.

Spec sheet

Nemotron 3 Nano 4B — 3.97B Dense.

Specifications
Parameters
3.97B
Architecture
Dense Transformer
Context Length
256K tokens
Capabilities
chat, coding, reasoning, math, tool_use
Release Date
2026-03-16
Provider
NVIDIA
Family
nemotron
VRAM requirements
QuantBPWVRAMQuality
Q4_K_M4.892.9 GB94%
Q5_K_S5.573.3 GB96%
Q5_K_M5.73.3 GB96%
Q6_K6.563.7 GB97%
Q8_08.54.7 GB100%
FP16168.4 GB100%
Benchmark scores
MMLU-PRO18.1
MATH95.4
IFEval88.0
BBH14.2
GPQA53.2
MUSR4.6
IFBench58.2
Terminal-Bench Hard6.8
τ²-Bench28.1
SciCode16.4
AA Long Context16.7
LiveCodeBench51.8
AIME78.5
GPQA Diamond51.3
HLE4.8
AA Intelligence14.7
AA Coding10.0
Terminal-Bench 2.13.7
Terminal-Bench 40.0
EQ-Bench 3588
Run command

Run Nemotron 3 Nano 4B locally with Ollama — needs 2.9 GB VRAM at Q4_K_M:

$ollama run nemotron-3-nano:4b
Compatible GPUs
30 @ Q4_K_M