KoSOLAR-v2.1: benchmark results

Provider: Other. Access: Open.

Unified ELO 1485 ± 18, rank #1222 of 2656 rated models, from 10 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open Korean LLM - Ko-GPQA-Diamond29.29Normalized accuracy (%)99.2
Open Korean LLM - Ko-Winogrande73.95Accuracy (%)99
Open Korean LLM - KorNAT-Helpful62.05Normalized accuracy (%)98.3
Open Korean LLM - KorNAT-Harmless72.22Normalized accuracy (%)95.3
Open Korean LLM - KorNAT-CKA29.24Normalized accuracy (%)79.1
Open Korean LLM Leaderboard39.53Average Score (%)72.6
Open Ko-LLM Leaderboard39.53Average (%)72.3
Open Korean LLM - Ko-EQ-Bench24.48EQ-Bench score70
Open Korean LLM - Ko-GSM8K (flexible extract)34.87Exact match, flexible extract (%)68.2
Open Korean LLM - Ko-IFEval29.07Strict accuracy, prompt/instruction mean (%)50.3

Interactive version: theaggregate.ai/model?slug=kosolar-v2-1 · How It Works · Data refreshed daily, snapshot 2026-09-19.