Specification

Release date: 10 Mar 2026
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Openness: open-weights
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
License: other
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Architecture: NemotronHForCausalLM
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Parameters: 123.6B
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
Active parameters: 12B
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Languages: en, fr, es, it, de, ja, zh
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
File size: 247.2 GB
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Hugging Face repo: nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
Pipeline tag: text-generation
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
Model card: huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
Downloads: 1,245,114
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 32 min agomedium
Likes: 424
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 32 min agomedium
Datasets: nvidia/nemotron-post-training-v3, nvidia/nemotron-pre-training-datasets
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Gated: No
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Hf inference providers: featherless-ai
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Last modified: 2026-08-25T15:48:50+00:00
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
Library name: transformers
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
License name: nvidia-nemotron-open-model-license
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
License url: https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-nemotron-open-model-license/
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Downloads all time: 5,627,017
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 32 min agomedium
Model type: nemotron_h
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Tags: transformers, safetensors, nemotron_h, text-generation, nvidia, pytorch, nemotron-3, latent-moe, mtp, custom_code, en, fr, es, it, de, ja, zh
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Weights dtype: BF16, F32
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium

Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →

Provenance

Attributed facts

24

Source tiers

T224

Freshest observation

32 min ago

Conflicts

None

Modalities

Modalities unavailable.

Capabilities

Tool calling
Unavailable
Structured output
Unavailable
Reasoning
Unavailable
Vision
Unavailable
Audio
Unavailable
Fine-tuning available
Unavailable

No capability flags have been observed from a source yet — we do not infer them.

Languages: en, fr, es, it, de, ja, zh
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium

No benchmark results recorded

Results appear when a tier 1–3 source publishes them; we never copy scores without a source.

Current prices

No current prices recorded

Prices appear when a provider publishes a public pricing page we crawl.

Price history

Estimated

Memory need = bytes per parameter (4-bit ≈ 0.5 × 1.15 overhead, 8-bit 1.0, fp16 2.0) + a KV-cache allowance. Not a measurement.

Estimated hardware fit
Hardware	Quantization	Memory	Est. need	Fits
NVIDIA DGX B200	4bit	1,440 GB	71.6 GB	Yes
Apple M3 Ultra	4bit	—	71.6 GB	Yes
Mac Studio (Apple M5 Ultra)	4bit	—	71.6 GB	Yes
AMD Instinct MI325X	4bit	256 GB	71.6 GB	Yes
AMD Instinct MI300X	4bit	192 GB	71.6 GB	Yes
Apple M2 Ultra	4bit	—	71.6 GB	Yes
NVIDIA B200	4bit	180 GB	71.6 GB	Yes
NVIDIA H200	4bit	141 GB	71.6 GB	Yes
NVIDIA H200 NVL	4bit	141 GB	71.6 GB	Yes
Apple M1 Ultra	4bit	—	71.6 GB	Yes
Apple M3 Max	4bit	—	71.6 GB	Yes
Apple M4 Max	4bit	—	71.6 GB	Yes
Mac Studio (Apple M5 Max)	4bit	—	71.6 GB	Yes
MacBook Pro (Apple M5 Max)	4bit	—	71.6 GB	Yes
NVIDIA DGX Spark	4bit	128 GB	71.6 GB	Yes
Apple M2 Max	4bit	—	71.6 GB	Yes
NVIDIA H100 NVL	4bit	94 GB	71.6 GB	Yes
NVIDIA A100 80GB	4bit	80 GB	71.6 GB	Yes
NVIDIA H100 SXM	4bit	80 GB	71.6 GB	Yes
Apple M1 Max	4bit	—	71.6 GB	No
Apple M4 Pro	4bit	—	71.6 GB	No
Mac mini (Apple M5 Pro)	4bit	—	71.6 GB	No
MacBook Pro (Apple M5 Pro)	4bit	—	71.6 GB	No
Apple M3 Pro	4bit	—	71.6 GB	No
Apple M1 Pro	4bit	—	71.6 GB	No
Apple M2 Pro	4bit	—	71.6 GB	No
Apple M4	4bit	—	71.6 GB	No
iMac (Apple M4)	4bit	—	71.6 GB	No
Mac mini (Apple M6)	4bit	—	71.6 GB	No
MacBook Air (Apple M5)	4bit	—	71.6 GB	No
MacBook Pro (Apple M5)	4bit	—	71.6 GB	No
NVIDIA GeForce RTX 5090	4bit	32 GB	71.6 GB	No
Apple M2	4bit	—	71.6 GB	No
Apple M3	4bit	—	71.6 GB	No
NVIDIA GeForce RTX 3090	4bit	24 GB	71.6 GB	No
NVIDIA GeForce RTX 4090	4bit	24 GB	71.6 GB	No
Apple M1	4bit	—	71.6 GB	No

No lineage recorded

Lineage comes from explicit `derived_from`, `fine_tuned_from`, `distilled_from` and `quantized_from` relations stated by sources.

Papers 2

arXiv:2512.20856Active
35
arXiv:2512.20848Active
35

Repositories 0

No repositories linked yet.

As of

Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.

Claim history · Architecture

1 claims · 1 propertiesShow all properties

Architecturearchitecture1

Claim history for Architecture
Value	Valid from → to	Status	Source	Confidence	Extractor
NemotronHForCausalLM	11 Sept 2026 21:55 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

11 Sept 2026
New modelModelNVIDIA-Nemotron-3-Super-120B-A12B-BF16NVIDIA
New model: NVIDIA-Nemotron-3-Super-120B-A12B-BF16 (NVIDIA)
huggingface
huggingface.co

Full timeline →

Source documents
Source	Document	Type	Tier	Last observed	Snapshots
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16	model_page	T2· Quality secondary	32 min ago	2
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/models?author=nvidia&p=0&sort=downloads	listing	T2· Quality secondary	1 h ago	3
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16/raw/main/README.md	model_card	T2· Quality secondary	1 h ago	1

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.