Benchmark Almanac

Nemotron Nano 12B v2 VL

Nemotron Nano 12B v2 VL places in the top three on HallusionBench.

NVIDIA Released October 28th, 2025 Family Nemotron Nano 2 The lab's model card Compare with others
50
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

124.2 #319 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • MMMU-Pro 28% #89 /90
  • SciCode 17.6% #49 /49
  • HallusionBench 73.1% #2 /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

BenchmarkScoreRankSource
HallusionBench 73.1% #2 of 31 Lab model card
MMLU-Pro 77.1% #100 of 167 Lab model card
POPE 87% #5 of 7 Lab model card
IFEval 78.2% #110 of 135 Lab model card
RULER 72.1% #37 of 42 Lab model card
MMBench-CN v1.1 76.3% #5 of 5 Lab model card

Math and science

BenchmarkScoreRankSource
MATH-500 96.9% #22 of 70 Lab model card
GPQA Diamond 64.1% #209 of 302 Lab model card
AIME 2025 71.3% #110 of 143 Lab model card

Coding

BenchmarkScoreRankSource
SciCode 17.6% #49 of 49 Lab model card

Agents and tools

BenchmarkScoreRankSource
ScreenSpot-Pro 5.50% #32 of 32 Lab model card
ScreenSpot-v2 42.8% #14 of 14 Lab model card
ScreenSpot 40.1% #11 of 11 Lab model card

Vision, audio and robotics

BenchmarkScoreRankSource
ChartQA 89.7% #7 of 37 Lab model card
DocVQA 94.4% #15 of 36 Lab model card
MathVista 76.9% #32 of 70 Lab model card
InfographicVQA 80.4% #13 of 26 Lab model card
OCRBench v2 61.2% #9 of 14 Official leaderboard
TextVQA 76.1% #11 of 17 Lab model card
AI2D 87.1% #34 of 52 Lab model card
MMStar 71.7% #24 of 36 Lab model card
OCRBench 85.6% #23 of 34 Lab model card
MMMU 68% #73 of 103 Lab model card
ZeroBench 18.3% #22 of 31 Lab model card
RealWorldQA 72.2% #28 of 38 Lab model card
MME-RealWorld 62.2% #3 of 4 Lab model card
LogicVista 58.4% #11 of 14 Lab model card
CVBench 78.3% #12 of 15 Lab model card
MMBench-EN v1.1 82.6% #19 of 23 Lab model card
BLINK 56.7% #24 of 29 Lab model card
MathVerse 58.2% #10 of 12 Lab model card
MTVQA-Test 24.3% #5 of 6 Lab model card
TreeBench 42.5% #6 of 7 Lab model card
MATH-Vision (MathVision) 53.6% #44 of 51 Lab model card
Video-MME 65.9% #46 of 51 Lab model card
MLVU 73.6% #21 of 22 Lab model card
CharXiv 41.3% #73 of 75 Lab model card
MMMU-Pro 28% #89 of 90 Lab model card
DynaMath 32.3% #23 of 23 Lab model card
MMLongBench-Doc 32.1% #19 of 19 Lab model card
We-Math 39.3% #13 of 13 Lab model card
MuirBench 44.2% #12 of 12 Lab model card
LongVideoBench 57% #11 of 11 Lab model card
CRPE 64.3% #4 of 4 Lab model card
OCR-Reasoning 36.4% #3 of 3 Lab model card
SEED-Bench-2-Plus 70% #3 of 3 Lab model card
VCR (Visual Caption Restoration) 71.8% #3 of 3 Lab model card
MMT-Bench 63.8% #2 of 2 Lab model card
R-Bench (real-world robustness) 69.7% #2 of 2 Lab model card
RD-TableBench 57.4% #1 of 1 Lab model card

Its lab's numbers put it at 124.2, number 319 of 404

The Self-Reported Index fits one score to the 38 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 121.2 to 127.2. How we compute it.

Its lab's number next to a leaderboard

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 62% OCRBenchv2 (English) 61.2% Leaderboard +0.8 pts

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 Nov 6, 2025

Capabilities