Comparison on Loong Set 2: 50K–100K Tokens
64.34LLM ScoreDisco-RAG
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Disco-RAGBase Model=Llama-3.3-70B2026.01 | 64.34 | 36 | |
| Disco-RAGBase Model=Qwen2.5-72B2026.01 | 64.06 | 30 | |
| StructRAGCondition=SOTA Results2026.01 | 63.71 | 36 | |
| Disco-RAGBase Model=Llama-3.1-8B2026.01 | 63.56 | 24 | |
| Llama-3.3-70BCondition=Standard RAG2026.01 | 53.37 | 22 | |
| RQ-RAGCondition=SOTA Results2026.01 | 50.83 | 16 | |
| Qwen2.5-72BCondition=Standard RAG2026.01 | 50.64 | 20 | |
| Llama-3.3-70BCondition=Full Context2026.01 | 47.93 | 26 | |
| Llama-3.1-8BCondition=Standard RAG2026.01 | 45.42 | 19 | |
| Qwen2.5-72BCondition=Full Context2026.01 | 44.47 | 25 | |
| Llama-3.1-8BCondition=Full Context2026.01 | 42.37 | 21 | |
| GraphRAGCondition=SOTA Results2026.01 | 14.29 | 0 |