Nemotron 3 Nano Omni 30B A3B

Nemotron 3 Nano Omni 30B A3B is a reasoning model from NVIDIA in the Nemotron 3 Nano Omni family. 19 benchmarks count toward its score, in 7 categories.

availableShows if the model has enough results for an index.
IndexOverall score out of 100.36.9 ±4.2
CoverageShare of the index weight with results.95%
SpeedOutput tokens per second.269/s
Input / 1MUS dollars per 1M input tokens.Free
Output / 1MUS dollars per 1M output tokens.Free
ContextMaximum tokens in one request.256K
EloLMArena rating and rank.N/A

The index is a score out of 100. The ± range shows how much it can change.

CapabilitiesScore per category, out of 100.

Out of 100
AgenticMulti-step tasks with tools.
33.5
CodingCode writing and repair.
35.0
ReasoningLogic problems and puzzles.
37.1
MultimodalTasks with images and text.
41.0
KnowledgeFacts and expert knowledge.
33.6
MultilingualTasks in many languages.
N/A
InstructionTasks with strict rules in the prompt.
52.8
MathMath problems.
41.7

Results

19 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result as a score out of 100.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
RefCOCO averageMultimodal90.5%RefCOCO dataset authors
AI2D test splitMultimodal88.5%Qwen
American Invitational Mathematics Examination 2025Math82.1%41.7Mathematical Association of America
Massive Multitask Language Understanding ProfessionalKnowledge77.3%42.3Yubo Wang et al.
CharXiv ReasoningMultimodal76.3%52.9CharXiv authors
Instruction Following BenchmarkInstruction74.2%52.4Benchmark authors
Video-MME without subtitleMultimodal72.2%Qwen
Graduate-Level Google-Proof Q&AKnowledge72.2%44.9David Rein et al.
GPQA DiamondKnowledge72.2%44.9David Rein et al.
Massive Multi-discipline Multimodal UnderstandingMultimodal70.8%40.3MMMU authors
LiveCodeBench v5Coding63.2%LiveCodeBench maintainers
Artificial Analysis IFBenchInstruction63.2%53.2Artificial Analysis
ScreenSpot ProMultimodal57.8%38.8Kaixin Li et al.
MMLongBench-DocMultimodal57.5%Qwen
Artificial Analysis MMMU-ProMultimodal53.2%32.1Artificial Analysis
OSWorldAgentic47.4%Z.AI
Artificial Analysis GPQA DiamondKnowledge46.9%16.6Artificial Analysis
τ²-Bench Tool-Agent-User EvaluationAgentic45.3%31.3Victor Barres et al.
Artificial Analysis Long Context ReasoningReasoning39.7%35.7Artificial Analysis
Scientific Code BenchmarkCoding32.0%41.4Benchmark authors
Artificial Analysis Omniscience AccuracyKnowledge15.2%32.7Artificial Analysis
Artificial Analysis Coding IndexCoding13.8%28.7Artificial Analysis
Artificial Analysis Intelligence IndexKnowledge10.3%35.2Artificial Analysis
Artificial Analysis Humanity's Last ExamKnowledge4.8%29.8Artificial Analysis
GDPval-AA normalizedAgentic0.0%35.6Artificial Analysis
Critical Physics TasksReasoning0.0%38.4Artificial Analysis

19 benchmarks count, from 20 of 26 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authors

More from NVIDIA

Nemotron 3 Ultra53.4Nemotron 3 Super 100B42.5Nemotron 3.5 Lightning 30B A3B NVFP437.4Nemotron 3 Nano 30B37.1Audio Flamingo 3 7BUnranked