Sina-Odin-7B-Merge: benchmark results
Azazelle's DARE-TIES merge, whose published recipe uses Mihaiii/Metis-0.3 as its base and rishiraj/smol-7b, SanjiWatsuki/openchat-3.5-1210-starling-slerp, Azazelle/Dumb-Maidlet as merge inputs. Its published Mistral configuration specifies 32 decoder layers. Provider: Azazelle. Access: Open.
Aggregate Rating 1290 ± 34, rank #8866 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 72.22 | Accuracy (%) (5-shot) | 28.6 |
| Open LLM Leaderboard v1 - GSM8K | 8.26 | Accuracy (%) (5-shot) | 28.2 |
| Open LLM Leaderboard v1 - ARC Challenge | 52.82 | Normalized accuracy (%) (25-shot) | 27.9 |
| Open LLM Leaderboard v1 - MMLU | 45.54 | Accuracy (%) (5-shot) | 24.1 |
| Open LLM Leaderboard v1 - HellaSwag | 68.86 | Normalized accuracy (%) (10-shot) | 19.6 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 39.2 | MC2 (%) (0-shot) | 10.4 |
Interactive version: theaggregate.ai/model?slug=sina-odin-7b-merge · How It Works · Data refreshed daily, snapshot 2026-11-08.