LMCocktail-10.7B-v1: benchmark results
Yhyu13's LM-Cocktail 10.7B, a 50/50 weight-space merge of Upstage's SOLAR 10.7B Instruct and rishiraj's meow model. Provider: Yhyu13. Released 2023-12-20. Access: Open.
Aggregate Rating 1458 ± 25, rank #3422 of 10916 rated models, from 18 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 71.03 | MC2 (%) (0-shot) | 91.1 |
| Open LLM Leaderboard v1 - HellaSwag | 88.13 | Normalized accuracy (%) (10-shot) | 89.6 |
| AlpacaEval | 84.78 | Length-Controlled Win Rate (%) | 89.5 |
| Open LLM Leaderboard v1 - WinoGrande | 83.35 | Accuracy (%) (5-shot) | 88.4 |
| Open Korean LLM - KorNAT-Harmless | 67.17 | Normalized accuracy (%) | 88.3 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 53.6 | Exact match, flexible extract (%) | 87.1 |
| Open LLM Leaderboard v1 - ARC Challenge | 70.65 | Normalized accuracy (%) (25-shot) | 86.6 |
| Open LLM Leaderboard v1 - MMLU | 66.21 | Accuracy (%) (5-shot) | 86.3 |
| AlpacaEval 1.0 | 92.22 | Win Rate (%) | 82.2 |
| Open Korean LLM - Ko-EQ-Bench | 31.36 | EQ-Bench score | 80.5 |
| Open LLM Leaderboard v1 - GSM8K | 64.97 | Accuracy (%) (5-shot) | 79.3 |
| Open Korean LLM - KorNAT-Helpful | 49.18 | Normalized accuracy (%) | 77.2 |
Interactive version: theaggregate.ai/model?slug=lmcocktail-10-7b-v1 · How It Works · Data refreshed daily, snapshot 2026-11-08.