Nemotron 3 Super 120B A12B

Nvidia llm

Mid-size Nemotron 3 model using a latent mixture-of-experts hybrid with multi-token prediction, targeting multi-agent applications over a 1M-token context.

overall ⌄

38.1

62 of 93 ranked

price · $/M tokens

self-hosted

open weights, no first-party API

to run it yourself ⌄

72 GB

minimum · 288 GB recommended

HF ↗ HuggingFace ↗

Benchmarks

What its maker published, and what anyone else measured. Every figure links the document it came from.

benchmark	published	measured
MMLU-Pro	83.73	—
AIME 2025	90.21	—
SWE-bench	60.47	—
HLE	18.26	—
BrowseComp	31.28	—

Lineage

No recorded lineage — root or standalone model.

view in full graph →

Metadata

org	Nvidia
released	2026-03-11
params	120B-A12B
license	nvidia-open-model
hf id	nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16