NVIDIA Nemotron Nano 9B V2: benchmark results
NVIDIA's 9B hybrid Mamba-2/Transformer text model with unified reasoning and non-reasoning modes. Provider: NVIDIA. Released 2025-08-18. Access: Open.
Aggregate Rating 1486 ± 15, rank #823 of 1656 rated models, from 353 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AGC-Bench - hypobench | 0.44 | Dataset z-score | 98.8 |
| AGC-Bench - grapheval_iclr | 1.73 | Dataset z-score | 97.6 |
| AGC-Bench - story_quality | 1.11 | Dataset z-score | 89.5 |
| ZeroEval MATH-500 | 97.8 | MATH-500 Score | 87.5 |
| AGC-Bench - pun_eval | 0.83 | Dataset z-score | 85.4 |
| EuroEval Italian Knowledge | 73.53 | Knowledge Average Score (%) | 84.3 |
| AA LiveCodeBench | 72.38 | Pass@1 (%) | 83.2 |
| EuroEval Swedish Knowledge | 71.67 | Knowledge Average Score (%) | 83.1 |
| AGC-Bench - story_generation_rocstories | 0.77 | Dataset z-score | 82.3 |
| EuroEval French Knowledge | 71.02 | Knowledge Average Score (%) | 82.2 |
| SpeechMap Compliance | 84 | % Requests Completed | 82.2 |
| EuroEval Spanish Knowledge | 71.78 | Knowledge Average Score (%) | 82 |
Interactive version: theaggregate.ai/model?slug=nvidia-nemotron-nano-9b-v2 · How It Works · Data refreshed daily, snapshot 2026-10-20.