dolphin-2.9.2-qwen2-7B: benchmark results
Cognitive Computations' dolphin-2.9.2-qwen2-7b, an uncensored fine-tune of Qwen2-7B trained at 16K sequence length on alignment-filtered data, with instruction, coding, agentic and function-calling skills. Provider: Cognitive Computations. Released 2024-06-06. Access: Open.
Aggregate Rating 1441 ± 20, rank #1047 of 1661 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-GSM8K (flexible extract) | 56.1 | Exact match, flexible extract (%) | 90.1 |
| Open Portuguese LLM - ASSIN2 RTE | 93.34 | Macro F1 (%) | 88.8 |
| Open Ko-LLM Leaderboard | 42.39 | Average (%) | 86.3 |
| Open Korean LLM Leaderboard | 42.37 | Average Score (%) | 86.3 |
| Open Portuguese LLM - BLUEX | 59.53 | Accuracy (%) | 76.5 |
| Open Portuguese LLM - OAB Exams | 49.93 | Accuracy (%) | 75.2 |
| Open LLM Leaderboard - MMLU-Pro | 40.51 | Score | 74.3 |
| Open Korean LLM - Ko-EQ-Bench | 23.31 | EQ-Bench score | 68.6 |
| Open Korean LLM - Ko-Winogrande | 62.12 | Accuracy (%) | 66.6 |
| Open Korean LLM - Ko-IFEval | 33.38 | Strict accuracy, prompt/instruction mean (%) | 66.3 |
| Open Korean LLM - KorNAT-CKA | 26.41 | Normalized accuracy (%) | 64.3 |
| Open LLM Leaderboard - MuSR | 41.91 | Score | 59.6 |
Interactive version: theaggregate.ai/model?slug=dolphin-2-9-2-qwen2-7b · How It Works · Data refreshed daily, snapshot 2026-10-23.