testcat-v0.4: benchmark results
Provider: Other. Access: Open.
Unified ELO 1457 ± 17, rank #1477 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 65 | Normalized accuracy (%) | 98.4 |
| Open Korean LLM - Ko-Winogrande | 72.77 | Accuracy (%) | 98.1 |
| Open Korean LLM - KorNAT-Harmless | 77.09 | Normalized accuracy (%) | 97.9 |
| Open Korean LLM - Ko-GPQA-Diamond | 28.28 | Normalized accuracy (%) | 96.6 |
| Open Korean LLM - KorNAT-CKA | 28.03 | Normalized accuracy (%) | 73.2 |
| Open Ko-LLM Leaderboard | 37.78 | Average (%) | 64.1 |
| Open Korean LLM Leaderboard | 37.78 | Average Score (%) | 64.1 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 21.83 | Exact match, flexible extract (%) | 54.1 |
| Open Korean LLM - Ko-EQ-Bench | 3.57 | EQ-Bench score | 49.6 |
| Open Korean LLM - Ko-IFEval | 25.26 | Strict accuracy, prompt/instruction mean (%) | 34 |
Interactive version: theaggregate.ai/model?slug=testcat-v0-4 · How It Works · Data refreshed daily, snapshot 2026-09-19.