Benchmark Almanac

Nemotron 3 Nano Omni 30B-A3B

Nemotron 3 Nano Omni 30B-A3B places in the top three on 2 benchmarks.

NVIDIA Released April 28th, 2026 Family Nemotron 3 The lab's model card Compare with others
34
Benchmarks with a score
0
Ranked first, out of five or more models
2
In the top three
NVIDIA Apr 2026

Nemotron 3 Nano Omni 30B-A3B

Open weights Small

136.7 #210 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • IFBench 74.2% #27 /80
  • SciCode 32% #43 /49
  • AA-LCR (Artificial Analysis Long Context Reasoning) 41% #33 /38

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
StudyArena Text 1,281 #10 of 85 Official leaderboard
IFBench 74.2% #27 of 80 Lab model card
MMLU-Pro 77.3% #97 of 167 Lab model card
AA-LCR (Artificial Analysis Long Context Reasoning) 41% #33 of 38 Lab model card

Math and science

BenchmarkScoreRankSource
GPQA Diamond 72.2% #165 of 302 Lab model card
AIME 2025 82.1% #85 of 143 Lab model card

Coding

BenchmarkScoreRankSource
LiveCodeBench v5 63.2% #22 of 47 Lab model card
SciCode 32% #43 of 49 Lab model card

Agents and tools

BenchmarkScoreRankSource
ScreenSpot 89.3% #4 of 11 Lab model card
ScreenSpot-v2 92.8% #7 of 14 Lab model card
ScreenSpot-Pro 57.8% #25 of 32 Lab model card
OSWorld-Verified 47.4% #37 of 41 Lab model card

Vision, audio and robotics

BenchmarkScoreRankSource
ChartQA 90.3% #5 of 37 Lab model card
DocVQA 95.6% #5 of 36 Lab model card
OCRBench v2 65.8% #3 of 14 Official leaderboard
MMLongBench-Doc 57.6% #5 of 17 Official leaderboard
InfographicVQA 86.8% #8 of 26 Lab model card
MathVista 82.8% #24 of 70 Lab model card
Daily-Omni 74.5% #5 of 14 Official leaderboard
VoiceBench 89.4% #5 of 12 Lab model card
WorldSense (omnimodal) 55.4% #5 of 12 Official leaderboard
TreeBench 51.6% #3 of 7 Lab model card
RefCOCO / RefCOCO+ / RefCOCOg 90.5% #13 of 29 Lab model card
TextVQA 81% #8 of 17 Lab model card
OCRBench 86.6% #17 of 34 Lab model card
AI2D 88.5% #27 of 52 Lab model card
MMMU 70.8% #58 of 103 Lab model card
CVBench 84% #9 of 15 Lab model card
OCR-Reasoning 54.1% #2 of 3 Lab model card
CharXiv (DQ) 88.9% #5 of 7 Lab model card
MMAU 74.6% #11 of 15 Lab model card
ParseBench 48.5% #34 of 44 Official leaderboard
CharXiv 63.6% #60 of 75 Lab model card
Video-MME 72.2% #42 of 51 Lab model card

Its lab's numbers put it at 136.7, number 210 of 404

The Self-Reported Index fits one score to the 27 capability results NVIDIA reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 135.1 to 138.4. How we compute it.

Its lab's numbers next to 5 leaderboards

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark Lab says Board says Gap
OCRBench v2 67% OCRBenchV2 (EN) 65.8% Leaderboard +1.2 pts
MMLongBench-Doc 57.5% MMlongBench Doc 57.6% Leaderboard, Omni; 3B activated (30B total) Same
VoiceBench 89.4% VoiceBench Avg, reasoning on (Table 8). Nine spoken subsets from 71.4 (SD-QA) to 100 (AdvBench) 89.4% Leaderboard Same
Daily-Omni 74.5% Daily Omni 74.5% Leaderboard, Omni-Modal Language Models (With Visual and Audio) Same
WorldSense (omnimodal) 55.4% World Sense (Video + Audio Q/A) 55.4% Leaderboard, modality A+V Same

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Its system card

Every result in the longer documents the lab published with it. 22 of them aren't in the launch post.

Nemotron 3 Nano Omni: Efficient and Open Multimodal Intelligence Apr 27, 2026

Capabilities