Nemotron 3 Super at a glance
Nemotron 3 Super has 120B parameters with 12B active, a 1M-token context window, and open weights, datasets and training recipes. NVIDIA says it delivers up to 5x the throughput of the previous Nemotron Super.
What changed with Nemotron 3 Super
On March 11, 2026, NVIDIA released Nemotron 3 Super, the middle model of the Nemotron 3 family it announced in December. It is a hybrid mixture-of-experts model with 120 billion total and 12 billion active parameters and a 1-million-token context window, aimed at subtasks inside multi-agent systems, such as coding agents that load a whole codebase or analysts that read long financial reports. NVIDIA says it reaches up to 5x higher throughput and up to 2x higher accuracy than the previous Nemotron Super; these are NVIDIA's claims.
The weights are released under a permissive license, together with more than 10 trillion tokens of training data, 15 reinforcement learning environments and evaluation recipes. The model is on Hugging Face, build.nvidia.com, Perplexity and OpenRouter, is packaged as an NVIDIA NIM microservice, and is offered through Google Cloud Vertex AI, Oracle Cloud Infrastructure and inference providers such as Baseten, DeepInfra and Fireworks AI.
Nemotron 3 Super in numbers
Values are copied from the linked source on the date shown. They are not IA Compa measurements.
- Context windowas of · Source for Context window: blogs.nvidia.com (opens in a new tab)
- 1M tokens
How it compares
Other models in the same category that we have covered, using each model's latest recorded value. (Models)
Where to compare Nemotron 3 Super
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
Why Nemotron 3 Super matters for businesses
An efficient open model with a long context lets agencies run high-volume agent steps on their own or a client's infrastructure, keeping data in-house and costs predictable, while saving premium closed models for the hardest steps.
Sources for this story
- New NVIDIA Nemotron 3 Super Delivers 5x Higher Throughput for Agentic AI (opens in a new tab)Official
NVIDIA Blog · blogs.nvidia.com
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #llm
- #nvidia
- #nemotron
- #open-weights
- #agents
