- MMMU* (val, ChatGPT judge) 68
- MathVista* 76.9
- AI2D (test) 87.1
- OCRBenchv2 (English) 62
- OCRBench 85.6
- OCR-Reasoning 36.4
- ChartQA (test) 89.7
- DocVQA (val) 94.4
- Video-MME w/o sub 65.9
Nemotron Nano 12B v2 VL
Nemotron Nano 12B v2 VL places in the top three on HallusionBench.
51
Benchmarks with a score
0
Ranked first, out of five or more models
1
In the top three
NVIDIA Oct 2025 Nemotron Nano 12B v2 VL
Open weights Small
No Epoch Capabilities Index score yet
- MMMU-Pro 28% #72 /72
- SciCode 17.6% #47 /47
- OCRBench v2 61.2% #9 /65
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 | |
|---|---|---|---|---|
| HallusionBench | 73.1% | #2 of 23 | Model card | |
| MMLU-Pro | 77.1% | #93 of 151 | Model card | |
| POPE | 87% | #5 of 7 | Model card | |
| IFEval | 78.2% | #102 of 126 | Model card | |
| RULER | 72.1% | #34 of 38 | Model card | |
| MMBench-CN v1.1 | 76.3% | #5 of 5 | Model card |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MATH-500 | 96.9% | #22 of 67 | Model card | |
| GPQA Diamond | 64.1% | #197 of 284 | Model card | |
| AIME 2025 | 71.3% | #110 of 143 | Model card |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench | 69.4% | #26 of 73 | Model card | |
| SciCode | 17.6% | #47 of 47 | Model card |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| ScreenSpot-Pro | 5.50% | #25 of 25 | Model card | |
| ScreenSpot-v2 | 42.8% | #14 of 14 | Model card | |
| ScreenSpot | 40.1% | #11 of 11 | Model card |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OCRBench v2 | 61.2% | #9 of 65 | Official leaderboard | |
| ChartQA | 89.7% | #7 of 37 | Model card | |
| DocVQA | 94.4% | #14 of 33 | Model card | |
| MathVista | 76.9% | #28 of 64 | Model card | |
| InfographicVQA | 80.4% | #12 of 22 | Model card | |
| TextVQA | 76.1% | #11 of 17 | Model card | |
| MMStar | 71.7% | #20 of 30 | Model card | |
| AI2D | 87.1% | #30 of 44 | Model card | |
| MMMU | 68% | #69 of 97 | Model card | |
| OCRBench | 85.6% | #20 of 28 | Model card | |
| ZeroBench | 18.3% | #18 of 25 | Model card | |
| RealWorldQA | 72.2% | #22 of 30 | Model card | |
| MME-RealWorld | 62.2% | #3 of 4 | Model card | |
| BLINK | 56.7% | #23 of 28 | Model card | |
| MMBench-EN v1.1 | 82.6% | #14 of 17 | Model card | |
| MathVerse | 58.2% | #10 of 12 | Model card | |
| MTVQA-Test | 24.3% | #5 of 6 | Model card | |
| MATH-Vision (MathVision) | 53.6% | #36 of 42 | Model card | |
| TreeBench | 42.5% | #6 of 7 | Model card | |
| Video-MME | 65.9% | #38 of 42 | Model card | |
| CVBench | 78.3% | #10 of 11 | Model card | |
| LogicVista | 58.4% | #11 of 12 | Model card | |
| CharXiv | 41.3% | #66 of 67 | Model card | |
| MMMU-Pro | 28% | #72 of 72 | Model card | |
| DynaMath | 32.3% | #17 of 17 | Model card | |
| MLVU | 73.6% | #16 of 16 | Model card | |
| MMLongBench-Doc | 32.1% | #13 of 13 | Model card | |
| MuirBench | 44.2% | #12 of 12 | Model card | |
| LongVideoBench | 57% | #11 of 11 | Model card | |
| We-Math | 39.3% | #11 of 11 | Model card | |
| CRPE | 64.3% | #4 of 4 | Model card | |
| OCR-Reasoning | 36.4% | #3 of 3 | Model card | |
| SEED-Bench-2-Plus | 70% | #3 of 3 | Model card | |
| VCR (Visual Caption Restoration) | 71.8% | #3 of 3 | Model card | |
| MMT-Bench | 63.8% | #2 of 2 | Model card | |
| R-Bench (real-world robustness) | 69.7% | #2 of 2 | Model card | |
| RD-TableBench | 57.4% | #1 of 1 | 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. 45 of them aren't in the launch post.
NVIDIA Nemotron Nano V2 VL 6 Nov 2025
Capabilities
- MMBench-EN v1.1 82.6
- MMBench-CN v1.1 76.3
- MMStar 71.7
- BLINK 56.7
- MuirBench 44.2
- HallusionBench 73.1
- ZeroBench 18.3
- CRPE 64.3
- POPE 87
- MME-RealWorld 62.2
- RealWorldQA 72.2
- MMT-Bench 63.8
- R-Bench (real-world robustness) 69.7
- WildVision 20.2
- MMMU 67.8
- MMMU-Pro 28
- MathVista 75.5
- MATH-Vision (MathVision) 53.6
- MathVerse 58.2
- DynaMath 32.3
- LogicVista 58.4
- We-Math 39.3
- MMLongBench-Doc 32.1
- OCRBench 83.5
- OCRBench v2 54.8
- ChartQA 84.9
- RD-TableBench 57.4
- AI2D 84.7
- TextVQA 76.1
- DocVQA 93.2
- InfographicVQA 80.4
- OCR-Reasoning 33.9
- VCR (Visual Caption Restoration) 71.8
- SEED-Bench-2-Plus 70
- CharXiv 41.3
- TreeBench 42.5
- CVBench 78.3
- ScreenSpot 40.1
- ScreenSpot-v2 42.8
- ScreenSpot-Pro 5.5
- LongVideoBench 57
- MLVU 73.6
- Video-MME 63
- MTVQA-Test 24.3
- MMMB
- Multilingual MMBench
- MATH-500 96.9
- AIME 2025 71.3
- GPQA Diamond 64.1
- LiveCodeBench 69.4
- IFEval 78.2
- SciCode 17.6
- MMLU-Pro 77.1
- RULER 72.1