NVIDIA
NVIDIA Nemotron 3 Nano is an open reasoning model optimized for fast, cost-efficient inference. Built with a hybrid MoE and Mamba architecture and trained on NVIDIA-curated synthetic reasoning data, it delivers strong multi-step reasoning with stable latency and predictable performance for agentic and production workloads.
NVIDIA model catalogRank #379 across 523
Rank #253 across 433
Rank #20 across 355
Percentile score by analysis domain.
* Cost is inverted: lower input, output, and blended prices rank higher.
Higher bars mean stronger relative placement.
| Metric | Domain | Value | Rank |
|---|---|---|---|
| Artificial Analysis Intelligence Index | overall | 13.2 | #379 |
| Artificial Analysis Coding Index | coding | 15.8 | #253 |
| Artificial Analysis Math Index | math | 13.3 | #226 |
| MMLU-Pro | reasoning | 57.9% | #273 |
| reasoning |
| 39.9% |
| #416 |
| Humanity's Last Exam | reasoning | 4.6% | #369 |
| LiveCodeBench | coding | 36.0% | #193 |
| SciCode | coding, reasoning | 23.0% | #357 |
| Output Speed | speed | 87.3 tok/s | #150 |
| Time to First Token | speed | 0.26s | #8 |
| Blended Price | cost | $0.088/M | #20 |
| Input Price | cost | $0.050/M | #23 |
| Output Price | cost | $0.200/M | #29 |
| Value Index | cost, overall | 150.0 | #37 |