mlx-optiq
← Models

NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16, quantized by OptiQ for Apple Silicon.

On disk3.1 GB
Memoryabout 5 GB
Capability Score66.68

Capability Score breakdown

The six metrics behind this model's Capability Score of 66.68, as reported in the model card. Values are compared with uniform 4-bit.

  • This model
  • Uniform 4-bit
Capability Score breakdown: NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bitMMLU: This model 64.0, Uniform 4-bit 63.3; GSM8K: This model 81.5, Uniform 4-bit 79.9; IFEval: This model 56.2, Uniform 4-bit 56.0; BFCL-V3: This model 94.0, Uniform 4-bit 94.0; HumanEval: This model 77.4, Uniform 4-bit 80.5; HashHop: This model 27.0, Uniform 4-bit 25.0MMLU64.063.3GSM8K81.579.9IFEval56.256.0BFCL-V394.094.0HumanEval77.480.5HashHop27.025.0

Use it

Install

pip install mlx-optiq

Serve it

An OpenAI- and Anthropic-compatible server on this Mac. Every app below can use it.

optiq serve --model mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

OptiQ Code

The terminal coding agent uses whatever model optiq serve is running.

optiq serve --model mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit
optiq code

OptiQ Lab

optiq lab --model mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

Claude Code

optiq run claude -m mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

Codex

optiq run codex -m mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

OpenCode

optiq run opencode -m mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

Mistral Vibe

optiq run vibe -m mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

Hermes Agent

optiq run hermes -m mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

OpenClaw

optiq run openclaw -m mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit

optiq run starts optiq serve if it is not running, sets the agent up for this session only, and stops the server when the agent exits.

mlx-lm, text only

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

FamilyNemotron · Nemotron docs
ArchitectureNemotron-H
Base modelnvidia/NVIDIA-Nemotron-3-Nano-4B-BF16
Size on disk3.1 GB
Memory neededabout 5 GB
Imagesno, text only
Capability Score66.68 (from the model card)
Downloads384 in the last month
Updated2026-09-14
Hugging Facemlx-community/NVIDIA-Nemotron-3-Nano-4B-OptiQ-4bit