Skip to content
llmwaves

Nemotron 3 Nano 30B A3B

NVIDIA · released Dec 14, 2025

Open weightsReasoningTool useStructured output

NVIDIA Nemotron 3 Nano 30B A3B is a small language MoE model with highest compute efficiency and accuracy for developers to build specialized agentic AI systems. The model is fully...

Specification

Context window
262K
Max output
262K
Knowledge cutoff
Not stated
Parameters
31.6B
Licence
other
Serving providers
3
Moderated
No
Uptime
100.0%

Intelligence

7.2

24th percentile

Coding

Coding Index

Agentic

2.0

Agentic Index

Output speed

138 t/s

Median across providers

Latency

353ms

Time to first token

Cost per task

$0.02

Estimated

Benchmarks

Where the score comes from

The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.

Evaluation scores

Percentage correct · higher is better

  • MMLU-Pro
    57.9%
  • GPQA Diamond
    39.9%
  • IFBench
    37.5%
  • LiveCodeBench
    36.0%
  • τ²-bench (Telecom)
    25.4%
  • SciCode
    23.0%
  • AIME 2025
    13.3%
  • Terminal-Bench Hard
    12.1%
  • AA-LCR (long context)
    8.7%
  • Humanity's Last Exam
    4.6%

An evaluation missing from this list was not run for this model — it is not a zero.

View as table
EvaluationScore
MMLU-Pro57.9%
GPQA Diamond39.9%
IFBench37.5%
LiveCodeBench36.0%
τ²-bench (Telecom)25.4%
SciCode23.0%
AIME 202513.3%
Terminal-Bench Hard12.1%
AA-LCR (long context)8.7%
Humanity's Last Exam4.6%

Against its peers

Intelligence Index · this model highlighted, nearest peers in grey

Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.

View as table
ModelIntelligence
Command A7.5
Nemotron 3 Nano 30B A3B7.2
Mistral Large 24077.0
Qwen2.5 Coder 32B Instruct6.9
GLM 4.5V6.8
GPT-46.8
Gemini 2.5 Flash Lite6.7
GPT-4o-mini6.7

Percentile among all indexed models

Intelligence24th
Agentic10th
Mathematics16th
Terminal work48th

Pricing

What it costs to run

List prices per million tokens, plus what one representative task works out to.

List price

Input / 1M tokens
$0.05
Output / 1M tokens
$0.2
Cached input / 1M
$0.03
Blended 3:1
$0.088

One task, estimated

$0.02

Input tokens
50,000
Output tokens
80,000
Profile
Reasoning

Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).

Serving providers

Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.