Llama 3.1 Nemotron Nano 4B v1.1 (Reasoning) — benchmark results

Provider: NVIDIA. Released 2025-04-07. Access: Open.

Unified ELO 1493 ± 62, rank #866 of 1841 rated models, from 12 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA MATH-50094.67Accuracy (%)79.9
AA LiveCodeBench49.31Pass@1 (%)55.6
AA AIME 202550Accuracy (%)48.1
AA Humanity's Last Exam5.1Accuracy (%)34.7
Artificial Analysis Intelligence Index8.54Intelligence Index27.5
AA MMLU-Pro55.6Accuracy (%)17.7
AA GPQA Diamond40.81Accuracy (%)17.3
Epoch AI - Scicode10.07Score9.6
AA TAU-2 Bench11.7Accuracy (%)9.1
AA-LCR0Score (self-reported)8.9
AA IFBench25.51Accuracy (%)8.5
AA Long Context Reasoning0Accuracy (%)6.7

Interactive version: theaggregate.ai/model?slug=llama-3-1-nemotron-nano-4b-v1-1-reasoning · How It Works · Data refreshed daily, snapshot 2026-07-25.