MiniCPM4.1-8B: benchmark results

Provider: OpenBMB. Access: Open.

Aggregate Rating 1454 ± 38, rank #3569 of 10916 rated models, from 13 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
JP-TL-Bench - Japanese to English - Easy3.86LT Score (0-10; Easy items only; anchored pairwise Bradley-T22.2
JP-TL-Bench - Japanese to English2.57LT Score (0-10; anchored pairwise Bradley-Terry vs 20 Base S20.6
JP-TL-Bench - Japanese to English - Hard1.79LT Score (0-10; Hard items only; anchored pairwise Bradley-T20.6
JP-TL-Bench - English to Japanese - Hard0.62LT Score (0-10; Hard items only; anchored pairwise Bradley-T14.3
JP-TL-Bench - English to Japanese0.54LT Score (0-10; anchored pairwise Bradley-Terry vs 20 Base S11.9
SuperCLUE General (September 2025) - Math Reasoning33.72Score11.1
JP-TL-Bench - English to Japanese - Easy0.35LT Score (0-10; Easy items only; anchored pairwise Bradley-T7.9
SuperCLUE General (September 2025) - Precise Instruction Following18.45Score4.2
SuperCLUE General (September 2025) - Agent29.72Score0
SuperCLUE General (September 2025) - Code Generation25.19Score0
SuperCLUE General (September 2025) - Hallucination Control48.09Score0
SuperCLUE General (September 2025) - Overall28.17Score0

Interactive version: theaggregate.ai/model?slug=minicpm4-1-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.