Nemotron 3 Super
Open SourceNVIDIA•Released 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.
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.”
“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.”
“The model powers the NVIDIA AI-Q research agent to the No. 1 position on DeepResearch Bench and DeepResearch Bench II leaderboards.”
Core Specs
Scenario Scores
Pros & Cons
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