Nemotron-Flash-1B: benchmark results

Provider: NVIDIA. Access: Open.

Aggregate Rating 1296 ± 37, rank #8677 of 10552 rated models, from 26 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Lara Mini Base Report (base-model protocol) - MATH (4-shot)44.9Math-Verify equivalence, Minerva prompts, 5,000 problems ove75.9
Lara Mini Base Report (base-model protocol) - MBPP+ (3-shot)45.2pass@1; must pass MBPP's own and the EvalPlus tests (%)42.6
Lara Mini Base Report (base-model protocol) - MMLU-Pro (5 (CoT)-shot)25Exact match (%), English benchmark40.7
Lara Mini Base Report (base-model protocol) - MGSM (8 (CoT)-shot)22Exact match on each language's native CoT answer phrase, ave35.2
Lara Mini Base Report (base-model protocol) - MBPP (3-shot)34.8pass@1 (execution) (%)33.3
Lara Mini Base Report (base-model protocol) - PIQA (0-shot)75.4Accuracy normalized by answer length (%), English benchmark33.3
Lara Mini Base Report (base-model protocol) - COPA (0-shot)79.8Accuracy (%), English benchmark32.4
Lara Mini Base Report (base-model protocol) - PAWS (English) (0-shot)60.2Accuracy (%), English benchmark25.9
Lara Mini Base Report (base-model protocol) - PAWS-X, All Languages (0-shot)52.2Accuracy (%) averaged over all languages/variants of the ben24.1
Lara Mini Base Report (base-model protocol) - XNLI (English) (0-shot)51.2Accuracy (%), English benchmark23.1
Lara Mini Base Report (base-model protocol) - GSM8K (5-shot)32.2Exact match (strict extraction: first "#### N", else first \22.2
Lara Mini Base Report (base-model protocol) - ARC-Challenge (25-shot)46Accuracy normalized by answer length (%), English benchmark20.4

Interactive version: theaggregate.ai/model?slug=nemotron-flash-1b · How It Works · Data refreshed daily, snapshot 2026-11-16.