NVIDIA: Nemotron 3 Super

● Excellent

by Nvidia Intelligent Agentic Open Weight Long Context Rank #51 of 75

8.1
/ 10

NVIDIA Nemotron 3 Super is a 120B-parameter open hybrid MoE model, activating just 12B parameters for maximum compute efficiency and accuracy in complex multi-agent applications. Built on a hybrid Mamba-Transformer...

Choose a model to compare against NVIDIA: Nemotron 3 Super

Specifications

Specifications for NVIDIA: Nemotron 3 Super
AttributeValue
Lab Nvidia
Tags Intelligent Agentic Open Weight Long Context
Overall Score 8.1/10
Release Date 2026-03
Context Window 1,000,000 tokens
Input Price / 1M $0.09
Output Price / 1M $0.45
Input Modalities Text
Output Modalities Text

Strengths

  • 120B MoE with only 12B active params for high efficiency
  • 1M token context window for very long tasks
  • Open-weight hybrid Mamba architecture
  • Strong multi-agent application performance

Weaknesses

  • Newer architecture with less ecosystem support
  • MoE routing can be inconsistent on edge cases
  • Benchmark coverage still growing

Best For

  • Complex multi-agent AI systems
  • Long-context reasoning and analysis
  • Efficient high-volume inference
  • Enterprise agentic deployments

Sources & Further Reading

Related Models

Scores are aggregated from public benchmarks (MMLU, HumanEval, MATH, GSM8K, LMSYS) and normalized to a 1–10 scale. Methodology →