NVIDIA-Nemotron-3-Super-120B-A12B

Home/Models/unsloth/NVIDIA-Nemotron-3-Super-120B-A12B

NVIDIA-Nemotron-3-Super-120B-A12B

unsloth/NVIDIA-Nemotron-3-Super-120B-A12B

kind: original
quant: bf16 (16-bit)
family: nemotron
modality: text-generation
root model: nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16
provider: unsloth
license: other
language: en, fr, es, it, de, ja, zh
context: 1M
library: —
vRAM (GB): —
size on disk: —
downloads: 470
likes: 3
created: —
last modified: 2026-03-11T16:13:46+00:00 (2026)
huggingface: https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Super-120B-A12B
tags: safetensorsnemotron_hnvidiapytorchnemotron-3latent-moemtptext-generationconversationalcustom_codeenfresitdejazhdataset:nvidia/nemotron-post-training-v3dataset:nvidia/nemotron-pre-training-datasetsarxiv:2512.20848arxiv:2512.20856base_model:nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16base_model:finetune:nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16license:other

Benchmark scores (17)

benchmark	category	score	context	conf	evidence
RULER	long-context	96.3	RULER @ 256k	medium	`\| RULER @ 256k \| 96.30 \| 96.74 \| 52.30 \|`
RULER	long-context	95.67	RULER @ 512k	medium	`\| RULER @ 512k \| 95.67 \| 95.95 \| 46.70 \|`
RULER	long-context	91.75	RULER @ 1M	medium	`\| RULER @ 1M \| 91.75 \| 91.33 \| 22.30 \|`
AIME 2025	math	90.21	AIME25 (no tools)	medium	`\| AIME25 (no tools) \| 90.21 \| 90.36 \| 92.50 \|`
MMLU-Pro	knowledge	83.73	MMLU-Pro	medium	`\| MMLU-Pro \| 83.73 \| 86.70 \| 81.00 \|`
GPQA	reasoning	82.7	GPQA (with tools)	medium	`\| GPQA (with tools) \| 82.70 \| — \| 80.09 \|`
LiveCodeBench	coding	81.19	LiveCodeBench (v5 2024-07↔2024-12)	medium	`\| LiveCodeBench (v5 2024-07↔2024-12) \| 81.19 \| 78.93 \| 88.00 \|`
MMLU-Pro	knowledge	79.36	MMLU-ProX (avg over langs)	medium	`\| MMLU-ProX (avg over langs) \| 79.36 \| 85.06 \| 76.59 \|`
GPQA	reasoning	79.23	GPQA (no tools)	medium	`\| GPQA (no tools) \| 79.23 \| 86.60 \| 80.10 \|`
Chatbot Arena	chat	73.88	Arena-Hard-V2	medium	`\| Arena-Hard-V2 \| 73.88 \| 75.15 \| 90.26 \|`
IFBench	instruction-following	72.56	IFBench (prompt)	medium	`\| IFBench (prompt) \| 72.56 \| 73.77 \| 68.32 \|`
SciCode	coding	42.05	SciCode (subtask)	medium	`\| SciCode (subtask) \| 42.05 \| 42.00 \| 39.00 \|`
BrowseComp	agentic	31.28	BrowseComp with Search	medium	`\| BrowseComp with Search \| 31.28 \| — \| 33.89 \|`
Terminal-Bench	agentic-coding	31.0	Terminal Bench Core 2.0	medium	`\| Terminal Bench Core 2.0 \| 31.00 \| 37.50 \| 18.70 \|`
Terminal-Bench	agentic-coding	25.78	Terminal Bench (hard subset)	medium	`\| Terminal Bench (hard subset) \| 25.78 \| 26.80 \| 24.00 \|`
Humanity's Last Exam	knowledge	22.82	HLE (with tools)	medium	`\| HLE (with tools) \| 22.82 \| — \| 19.0 \|`
Humanity's Last Exam	knowledge	18.26	HLE (no tools)	medium	`\| HLE (no tools) \| 18.26 \| 25.30 \| 14.90 \|`

Discussions (0)

none scraped.