Skip to content
N

Nemotron 3 Super

Open Source

NVIDIAReleased on 2026-03-13

NVIDIA's flagship open-source model for agentic AI, featuring 120B total parameters with 12B active (MoE). Hybrid Mamba-Transformer architecture delivers 5x throughput vs previous Nemotron Super. 1M context window prevents goal drift in complex multi-agent workflows. #1 on DeepResearch Bench.

82
Overall Score

Voice of the community

sample 45

Running really well on my Macbook Pro M3 Max 128gb at a Q4 quant and the 1M context window. Running it through some of my LLM games it handles the specific output formats really well and the writing quality seems solid.

Reddit r/LocalLLaMA2026-03-12

Nemotron 3 Super has set new standards, claiming the top spot on Artificial Analysis for efficiency and openness with leading accuracy among models of the same size.

NVIDIA Blog2026-03-13

The model powers the NVIDIA AI-Q research agent to the No. 1 position on DeepResearch Bench and DeepResearch Bench II leaderboards.

NVIDIA Blog2026-03-13

Core Specs

1000K
Context Window
40K
Max Output
ReasoningOpen Sourcetext

Pros & Cons

Sentiment85% +15% ·0% −

Pros

  • +1M context window for full workflow state
  • +5x throughput vs previous Nemotron Super
  • +Open weights under permissive license
  • +#1 on DeepResearch Bench I & II
  • +Multi-token prediction for 3x faster inference

Cons

  • Text-only (no multimodal support)
  • Requires high-end hardware for self-hosting
  • New release, limited community feedback

Pricing

Input (per 1M tokens)$0.40
Output (per 1M tokens)$2.20
Free trial available
Updated on 2026-03-13

Get Started

1Visit the provider's website
2Create an account
3Start using the model

Benchmarks

deepResearchBench1%
artificialAnalysisEfficiency1%

Reliability

Incidents (30d)0
Open-source, reliability depends on hosting provider