NVIDIA-Nemotron-Nano-9B-v2-Japanese (Thinking): benchmark results

Provider: NVIDIA. Released 2026-02-17. Access: Open.

Unified ELO 1570 ± 1, rank #837 of 3078 rated models, from 101 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Nejumi 4 - BFCL - Live AST74.07Accuracy (%)89.3
Swallow - English MT-Bench - Math98.7Judge Score (normalized, %)81.3
Nejumi 4 - Toxicity - Violation Categories48.41Criteria met (%)78.3
Swallow - Japanese MT-Bench - Math97.8Judge Score (normalized, %)76.9
Nejumi 4 - HalluLens96Hallucination resistance (%)75.4
Nejumi 4 - BFCL - Irrelevance Detection86.67Accuracy (%)68.8
Nejumi 4 - jaster (0-shot) - JSICK81Exact match (%)68
Swallow - English MT-Bench - Writing73.7Judge Score (normalized, %)66.4
Swallow - English MT-Bench - Reasoning83.5Judge Score (normalized, %)62.7
Swallow - Post-trained Japanese - PolyMath High and Top39.6Accuracy (%)59.7
Swallow - Post-trained English - IFBench43Instruction-Level Strict Accuracy (%)59
Swallow - Japanese MT-Bench - Reasoning73.8Judge Score (normalized, %)56.7

Interactive version: theaggregate.ai/model?slug=nvidia-nemotron-nano-9b-v2-japanese-thinking · How It Works · Data refreshed daily, snapshot 2026-09-19.