- CVBench2D 84
- OCRBenchV2 (EN) 67
- OSWorld 47.4
- Charxiv Reasoning 63.6
- MMlongBench Doc 57.5
- MathVista_MINI 82.8
- OCR_Reasoning 54.1
- Video MME 72.2
- World Sense (Video + Audio Q/A) 55.4
- Daily Omni 74.5
- Voice interaction (Speech Instruction Following) 89.4
- MMAU (BF16, non-reasoning, quantization table) 74.6
Nemotron 3 Nano Omni 30B-A3B
Nemotron 3 Nano Omni 30B-A3B places in the top three on 2 benchmarks.
33
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
No Epoch Capabilities Index score yet
- IFBench 74.2% #26 /70
- SciCode 32% #43 /47
- AA-LCR (Artificial Analysis Long Context Reasoning) 41% #28 /31
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
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| IFBench | 74.2% | #26 of 70 | Model card | |
| MMLU-Pro | 77.3% | #91 of 151 | Model card | |
| AA-LCR (Artificial Analysis Long Context Reasoning) | 41% | #28 of 31 | Model card |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GPQA Diamond | 72.2% | #155 of 284 | Model card | |
| AIME 2025 | 82.1% | #85 of 143 | Model card |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench v5 | 63.2% | #21 of 44 | Model card | |
| SciCode | 32% | #43 of 47 | Model card |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| ScreenSpot | 89.3% | #4 of 11 | Model card | |
| ScreenSpot-v2 | 92.8% | #7 of 14 | Model card | |
| ScreenSpot-Pro | 57.8% | #20 of 25 | Model card | |
| OSWorld-Verified | 47.4% | #35 of 37 | Model card |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OCRBench v2 | 65.8% | #3 of 65 | Official leaderboard | |
| ChartQA | 90.3% | #5 of 37 | Model card | |
| DocVQA | 95.6% | #5 of 33 | Model card | |
| MathVista | 82.8% | #20 of 64 | Model card | |
| InfographicVQA | 86.8% | #8 of 22 | Model card | |
| TreeBench | 51.6% | #3 of 7 | Model card | |
| TextVQA | 81% | #8 of 17 | Model card | |
| OCRBench | 86.6% | #14 of 28 | Model card | |
| VoiceBench | 89.4% | #5 of 10 | Model card | |
| AI2D | 88.5% | #23 of 44 | Model card | |
| MMLongBench-Doc | 57.5% | #7 of 13 | Model card | |
| MMMU | 70.8% | #54 of 97 | Model card | |
| RefCOCO / RefCOCO+ / RefCOCOg | 90.5% | #11 of 19 | Model card | |
| CVBench | 84% | #7 of 11 | Model card | |
| OCR-Reasoning | 54.1% | #2 of 3 | Model card | |
| MMAU | 74.6% | #10 of 14 | Model card | |
| CharXiv (DQ) | 88.9% | #5 of 7 | Model card | |
| ParseBench | 48.5% | #33 of 42 | Official leaderboard | |
| WorldSense (omnimodal) | 55.4% | #4 of 5 | Model card | |
| CharXiv | 63.6% | #54 of 67 | Model card | |
| Video-MME | 72.2% | #34 of 42 | Model card | |
| Daily-Omni | 74.5% | #6 of 6 | Model card |
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.
Capabilities
- MMMU 70.8
- MathVista 82.8
- MMLongBench-Doc 57.5
- OCRBench 86.6
- OCRBench v2 67
- ChartQA 90.3
- DocVQA 95.6
- AI2D 88.5
- TextVQA 81
- InfographicVQA 86.8
- OCR-Reasoning 54.1
- CharXiv 63.6
- CharXiv (DQ) 88.9
- TreeBench 51.6
- CVBench 84
- RefCOCO / RefCOCO+ / RefCOCOg 90.5
- ScreenSpot 89.3
- ScreenSpot-v2 92.8
- ScreenSpot-Pro 57.8
- OSWorld-Verified 47.4
- Video-MME 72.2
- Open ASR Leaderboard 5.95
- MMAU 74.6
- VoiceBench 89.4
- Daily-Omni 74.1
- WorldSense (omnimodal) 55.4
- MMLU-Pro 77.3
- GPQA Diamond 72.2
- LiveCodeBench v5 63.2
- AIME 2025 82.1
- IFBench 74.2
- AA-LCR (Artificial Analysis Long Context Reasoning) 41
- tau2-bench
- SciCode 32