Nova 2.0 Pro Preview (Low): benchmark results

Nova 2.0 Pro Preview evaluated at the low reasoning-effort setting. Provider: Amazon. Released 2025-08-01. Access: API.

Unified ELO 1586 ± 1, rank #491 of 1761 rated models, from 21 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA IFBench79.59Accuracy (%)97.8
AA TAU-2 Bench90.64Accuracy (%)85.4
AA Long Context Reasoning64.67Accuracy (%)60.2
AA Omniscience - Health23Accuracy (%)59.9
BenchmarkList ECI118Capability Index (ECI)59.3
AA Omniscience - Humanities & Social Sciences22.3Accuracy (%)57.4
AA GPQA Diamond75.05Accuracy (%)56.9
AA Omniscience - Law13Accuracy (%)55.8
AA Terminal-Bench Hard17.42Accuracy (%)55.7
Artificial Analysis Intelligence Index13.35Intelligence Index54.9
AA Omniscience - Software Engineering (SWE)28.1Accuracy (%)54
AA Omniscience-40.53Score47.6

Interactive version: theaggregate.ai/model?slug=nova-2-0-pro-preview-low · How It Works · Data refreshed daily, snapshot 2026-09-05.