Nova Micro: benchmark results
Amazon's Nova Micro, the smallest, text-only, lowest-cost first-generation Nova model (December 2024). Provider: Amazon. Released 2024-12-03. Access: API.
Aggregate Rating 1467 ± 17, rank #898 of 1656 rated models, from 168 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenRouter Tau2-Bench Airline | 78 | Accuracy (%) | 93.4 |
| HELM NaturalQuestions (Open) | 77.86 | F1 (%) | 91.1 |
| HELM Lite - LegalBench - International Citizenship Questions | 62.9 | Exact Match (%) | 88.9 |
| HELM Lite - MATH Level 1 - Geometry | 71.05 | Equivalent (CoT) (%) | 87.2 |
| HELM Lite - MATH Level 1 - Precalculus | 77.19 | Equivalent (CoT) (%) | 82.2 |
| OpenRouter GPQA Diamond | 89 | Accuracy (%) | 76.6 |
| HELM Lite - MATH Level 1 - Number Theory | 80 | Equivalent (CoT) (%) | 75 |
| HELM Lite - MATH Level 1 | 76 | Equivalent (CoT) (%) | 72.2 |
| HELM Lite - MATH Level 1 - Prealgebra | 89.53 | Equivalent (CoT) (%) | 72.2 |
| AA-Omniscience Hallucination Rate | 65.08 | Hallucination Rate (%) | 66.8 |
| HELM Lite - WMT 2014 (de-en) | 19.73 | BLEU-4 (%) | 66.7 |
| HELM Lite - LegalBench - Function of Decision Section | 41.69 | Exact Match (%) | 65.6 |
Interactive version: theaggregate.ai/model?slug=nova-micro · How It Works · Data refreshed daily, snapshot 2026-10-20.