Nemotron 3 Super
NVIDIA · 1 configuration
Intelligence12.8
Input / 1M$0.30
Output / 1M$0.90
Context1M
Tariff: NVIDIA · Source: artificial-analysis · USD per million tokens
Thinking configuration
What does more thinking buy you?
Compare this model’s measured thinking configurations.
Scroll the chart ↔ · tap a point for details
Efficient frontier: best measured score at each cost1 measured configurations · 0 without both values
Selected: thinking thinking · Click a point to switch configuration
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...
Intelligence IndexArtificial Analysis composite index, not a percentage. Different versions of the index are not directly comparable.
12.8 pointsGPQA DiamondGraduate-level scientific questions. Measures specialist reasoning rather than web-search quality.
80.0 %Humanity’s Last ExamDifficult questions across many disciplines, compared under the same evaluation setup.
20.8 %SciCodeCode generation for scientific research problems.
36.2 %Terminal-Bench 2.1Agent tasks in a terminal environment. Results also depend on the evaluation harness.
38.6 %τ²-BenchTool use and interaction in assistance scenarios. Does not replace an evaluation of your integration.
67.8 %IFBenchVerifiable instruction following. A proxy for precision rather than a benchmark of creative style.
71.5 %Omniscience accuracyFactual accuracy in Artificial Analysis's Omniscience evaluation.
24.3 %Specifications from the catalog
Inputtext
Outputtext
Tool callingYes
Structured outputYes
Speed164.9 tok/s
First token1.30 s
Maximum output235.9K
Catalog listing date2026-03-11