llmboard.ai
Benchmarks
CompareRankings
llmboard.ai
Benchmarks
CompareRankings
HomeModelsNemotron 3 Super

NVIDIA model product

Nemotron 3 Super

Nemotron 3 Super is a 120B total / 12B active parameter hybrid Mamba-Attention Mixture-of-Experts model optimized for agentic reasoning, coding, planning, tool calling, and long-context analysis. It introduces LatentMoE (projecting tokens into a compressed latent space for expert routing, enabling 4x more experts at the same inference cost), Multi-Token Prediction for native speculative decoding (up to 3x faster generation), and native NVFP4 pretraining on Blackwell. The hybrid architecture interleaves Mamba-2 layers for linear-time sequence processing with strategically placed Transformer attention layers as global anchors, supporting a 1M-token context window. Pre-trained on 25 trillion tokens and post-trained with multi-environment RL across 21 configurations using NeMo Gym/RL with 1.2 million rollouts. Achieves up to 5x higher throughput than previous Nemotron Super and 2.2x higher throughput than GPT-OSS-120B while maintaining comparable accuracy.

Updated Aug 10, 2026. Default version: Nemotron 3 Super (120B A12B)

Compare
LLMBoard score59.1Nemotron 3 Super (120B A12B)
Coverage100%10 benchmark families
Context window262.1KTokens
Official input priceN/AOfficial price unavailable

On this page

  • Specification
  • Capability
  • Benchmarks
  • Arena
  • Pricing
  • Versions
  • About
  • Compare
  • Similar models
  • FAQ

Model specification

Structured fields from the published default version.

Version
Nemotron 3 Super (120B A12B)
Released
Mar 11, 2026
Knowledge cutoff
Jun 1, 2025
Parameters
120B
Context window
262.1K
Max output
262.1K
Inputs
text
Outputs
text
Open weights
Yes
License
NVIDIA Open Model License Agreement

Capability profile

This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.

Nemotron 3 Super (120B A12B) category scores

Benchmark results

Published benchmark records for the scored version Nemotron 3 Super (120B A12B).

23 rows
Columns

Show columns

WMT24++0.9123100.0%CAug 7, 2026
RULER0.92466.7%CAug 7, 2026
Bird-SQL (dev)0.45733.3%CAug 7, 2026
Arena-Hard v20.761666.7%CAug 7, 2026
LiveCodeBench0.877391.7%CAug 7, 2026
HMMT 20250.983378.1%CAug 7, 2026
SciCode0.4101847.1%CAug 7, 2026
AA-LCR0.6121521.4%CAug 7, 2026
MMLU-ProX0.8123264.5%CAug 7, 2026
Multi-Challenge0.6122960.7%CAug 7, 2026
IFBench0.7132855.6%CAug 7, 2026
Tau2 Airline0.6182322.7%CAug 7, 2026
Terminal-Bench0.3222512.5%CAug 7, 2026
Tau2 Retail0.624268.0%CAug 7, 2026
MMLU-Pro0.82912978.1%CAug 7, 2026
Tau2 Telecom0.6293517.6%CAug 7, 2026
SWE-bench Multilingual0.532346.1%CAug 7, 2026
AIME 20250.94711459.3%CAug 7, 2026
Terminal-Bench 2.00.349490.0%CAug 7, 2026
Humanity's Last Exam0.2539242.9%CAug 7, 2026
BrowseComp0.354587.0%CAug 7, 2026
GPQA0.87023370.3%CAug 7, 2026
SWE-Bench Verified0.58610417.5%CAug 7, 2026

Arena results

Preference and agent-evaluation signals from published Arena datasets.

No published Arena match

The default version has no published Arena rows, or its source alias has not been resolved.

Pricing

Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.

Official API
N/A
Official provider
N/A
Lowest third-party
N/A
Tracked offerings
1
1 rows
Columns

Show columns

Kenarinemotron-3-super-120b-a12bglobalN/AN/A262.1KAug 7, 2026

Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.

Nemotron 3 Super versions

All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.

1 rows
Columns

Show columns

Nemotron 3 Super (120B A12B)Mar 11, 202659.1120B262.1K262.1KYesNVIDIA Open Model License Agreement

What is Nemotron 3 Super?

A concise description based on the published model registry.

Nemotron 3 Super is a 120B total / 12B active parameter hybrid Mamba-Attention Mixture-of-Experts model optimized for agentic reasoning, coding, planning, tool calling, and long-context analysis. It introduces LatentMoE (projecting tokens into a compressed latent space for expert routing, enabling 4x more experts at the same inference cost), Multi-Token Prediction for native speculative decoding (up to 3x faster generation), and native NVFP4 pretraining on Blackwell. The hybrid architecture interleaves Mamba-2 layers for linear-time sequence processing with strategically placed Transformer attention layers as global anchors, supporting a 1M-token context window. Pre-trained on 25 trillion tokens and post-trained with multi-environment RL across 21 configurations using NeMo Gym/RL with 1.2 million rollouts. Achieves up to 5x higher throughput than previous Nemotron Super and 2.2x higher throughput than GPT-OSS-120B while maintaining comparable accuracy.

Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.

Data snapshot: 2026-08-07. Editorial model content is not available in the backend.

Nemotron 3 Super vs nearby models

Open a comparison with the three ranked models immediately above and below this model.

Nemotron 3 SupervsLongCat Flash ThinkingNemotron 3 SupervsNemotron Nano 9BNemotron 3 SupervsKimi K2 ThinkingNemotron 3 SupervsClaude Opus 3Nemotron 3 SupervsGPT-4.5Nemotron 3 SupervsQwen3 VL 235B A22B Thinking

Models similar to Nemotron 3 Super

Recommendations prioritize the same model type and family, then the closest published LLMBoard score.

#52+2.1
NV

Nemotron Nano 9B

NVIDIA

61.2 LLMBoard

DetailsCompare
#47+3.5
NV

Llama 3.1 Nemotron Ultra 253B

NVIDIA

62.6 LLMBoard

DetailsCompare
#46+3.5
NV

Llama 3.3 Nemotron Super 49B

NVIDIA

62.6 LLMBoard

DetailsCompare
#27+13.1
NV

Nemotron 3 Ultra

NVIDIA

72.1 LLMBoard

DetailsCompare
#97-14.0
NV

Nemotron 3 Nano

NVIDIA

45.1 LLMBoard

DetailsCompare
#100-14.9
NV

Llama 3.1 Nemotron Nano 8B

NVIDIA

44.2 LLMBoard

DetailsCompare

FAQ

Common questions about Nemotron 3 Super.

When was Nemotron 3 Super released?

Nemotron 3 Super's default version was released on Mar 11, 2026.

How much does Nemotron 3 Super cost?

No official standard PAYG price is currently available for Nemotron 3 Super.

Who created Nemotron 3 Super?

Nemotron 3 Super is published under NVIDIA in the model registry.

What is the context window for Nemotron 3 Super?

The default version has a 262.1K token context window.

Is Nemotron 3 Super open weight?

Yes. The default version is marked as open weight under NVIDIA Open Model License Agreement .

How many API providers offer Nemotron 3 Super?

1 published provider offerings are linked to the default version.

What models should I compare Nemotron 3 Super with?

Nearby ranked alternatives include LongCat Flash Thinking, Nemotron Nano 9B, Kimi K2 Thinking.

Rankings

OverallCodingText ArenaPricing

Modalities

Image GenerationVideo GenerationSpeech-to-TextEmbeddings

Benchmarks

All BenchmarksReasoningMathCoding

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai