Sarashina2.2-V-3B: benchmark results

Provider: Other.

Aggregate Rating 1499 ± 48, rank #2366 of 10916 rated models, from 10 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
JMMMU-Pro - Culture Specific54Accuracy (%)84.6
JMMMU-Pro - Japanese Art48Accuracy (%)84.6
JMMMU-Pro - Japanese Heritage51.33Accuracy (%)84.6
JMMMU-Pro - Japanese History51.33Accuracy (%)84.6
JMMMU-Pro42.88Overall (percent)69.2
JMMMU-Pro - Culture Agnostic36.25Accuracy (%)69.2
JMMMU-Pro - Overall42.88Accuracy (%)69.2
JMMMU-Pro - World History65.33Accuracy (%)69.2
HakushoBench (Direct)37.7Accuracy (%), direct prompt (answer accurately and concisely28.6
HakushoBench (CoT)32.3Accuracy (%), chain-of-thought prompt (think step by step, t0

Interactive version: theaggregate.ai/model?slug=sarashina2-2-v-3b · How It Works · Data refreshed daily, snapshot 2026-11-08.