Long Context Understanding on Ruler (Accuracy Metric)
93.2AccuracyMinistral-3-8B
Evaluation results
| Method | Links | |
|---|---|---|
| Ministral-3-8BParameters=8B2026.05 | 93.2 | |
| Qwen3-8BParameters=8B2026.05 | 92.6 | |
| Llama-3.1-8BParameters=8B2026.05 | 91.9 | |
| Qwen3-4BParameters=4B2026.05 | 90.9 | |
| GPT-5 nanoScale=nano2026.05 | 89.2 | |
| gemma-3-12b-itParameters=12B2026.05 | 87.5 | |
| Llama-3.2-3BParameters=3B2026.05 | 84.5 | |
| gpt-oss-20bParameters=20B2026.05 | 78 | |
| gemma-3-4b-itParameters=4B2026.05 | 73.9 | |
| Moonlight-16B-A3BParameters=16B2026.05 | 60.2 | |
| EngGPT2-16B-A3BParameters=16B2026.05 | 60 | |
| LLaMAntino-3-8BParameters=8B2026.05 | 56.1 | |
| FastwebMIIA-7BParameters=7B2026.05 | 55.9 | |
| Velvet-14BParameters=14B2026.05 | 40 | |
| deepseek-moe-16bParameters=16B2026.05 | 39.9 | |
| Minerva-7BParameters=7B2026.05 | 27.1 |