Multi-Hop Question Answering on Multi-Hop QA
73.1AccuracyHippoRAG2
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HippoRAG2Retrieval Strategy=Text-only Graph-based RAG, Input Modality=Text-only2026.06 | 73.1 | — | |
| KG4VDRetrieval Strategy=Text-only Graph-based RAG, Input Modality=Text-only2026.06 | 72.5 | — | |
| LightRAGRetrieval Strategy=Text-only Graph-based RAG, Input Modality=Text-only2026.06 | 60.5 | — | |
| TARSEBackbone=14B2026.03 | 48.5 | — | |
| NaiveRAGRetrieval Strategy=Embedding RAG, Input Modality=Text-only2026.06 | 48.1 | — | |
| GraphRAGRetrieval Strategy=Text-only Graph-based RAG, Input Modality=Text-only2026.06 | 46.3 | — | |
| Claude-Sonnet-4-52026.01 | 45.1 | 85 | |
| TARSEBackbone=7B2026.03 | 45.1 | — | |
| Gemini-2.5-Flash2026.01 | 42.9 | 73 | |
| Qwen3-235B-A22B-Instruct2026.01 | 40.6 | 83.2 | |
| GPT-4.1-MINI2026.01 | 40.2 | 83.6 | |
| No DocumentsRetrieval Strategy=Without Retrieval, Input Modality=Text-only2026.06 | 38.2 | — | |
| i-MedRAG-Qwen2.5Backbone=7B2026.03 | 35.6 | — | |
| GLM-Z1-32B2026.01 | 35.3 | 66.5 | |
| Meta-Llama-3.1-70B-Instruct2026.01 | 35.2 | 78.9 | |
| Grok-4-Fast2026.01 | 34.9 | 74.5 | |
| Qwen3-32B2026.01 | 34.2 | 73.9 | |
| rStar-Qwen2.5Backbone=14B2026.03 | 34.1 | — | |
| Deepseek-V3.12026.01 | 33.5 | 77.4 | |
| MedRAG-Qwen2.5Backbone=7B2026.03 | 33.2 | — | |
| DeepSeek-R1-Distill-Qwen-32B2026.01 | 32.8 | 62.4 | |
| Qwen2.5-InstructBackbone=14B2026.03 | 32.5 | — | |
| Qwen2.5-72B-Instruct2026.01 | 32.2 | 80.1 | |
| QwQ-32B2026.01 | 31.2 | 74.1 | |
| rStar-Qwen2.5Backbone=7B2026.03 | 29.1 | — | |
| Qwen2.5-14B-Instruct2026.01 | 28.7 | 70.6 | |
| DeepSeek-R1-Distill-Qwen-14B2026.01 | 28.5 | 67 | |
| Qwen2.5-32B-Instruct2026.01 | 26.8 | 76.4 | |
| Qwen2.5-InstructBackbone=7B2026.03 | 26.3 | — | |
| Meta-Llama-3.1-8B-Instruct2026.01 | 19.6 | 69.8 | |
| Qwen2.5-7B-Instruct2026.01 | 18.8 | 58.4 | |
| DeepSeek-R1-Distill-Llama-8B2026.01 | 18.2 | 58.4 | |
| Qwen2.5-1.5B-Instruct2026.01 | 17.7 | 65.6 | |
| OpenReasoning-Nemotron-14B2026.01 | 13.2 | 65.4 | |
| DeepSeek-R1-Distill-Qwen-7B2026.01 | 13 | 58.3 |