Method Drift›Long-context / context-window extension
RAG
Long-context / context-window extension
superseded — cited as a baseline and beaten by newer methods
2 papers critique it · 2 beat it on benchmarks
What papers say
Verbatim critique sentences, each from a paper that cites RAG as a baseline.
“RAG's filtering mechanism is inherently incapable of exploiting interdependencies across the entire context, limiting its effectiveness on tasks that require complex reasoning.”
— Graph of Agents: Principled Long Context Modeling by Emergent Multi-Agent Collaboration“RAG ... can mitigate the challenges mentioned above but introduce issues such as information loss for inappropriate extraction”
— Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
Beaten on benchmarks
Head-to-head results where a newer method reports beating RAG. Values are copied from the source paper's tables — verify against the cited paper.
- Graph of Agents: Principled Long Context Modeling by Emergent Multi-Agent Collaboration
GoA beats RAG · Average F1 [Llama 3.1 8B, 2K context]
48.67 vs 43.18
- Graph of Agents: Principled Long Context Modeling by Emergent Multi-Agent Collaboration
GoA beats RAG · Average F1 [Llama 3.1 8B, 8K context]
49.18 vs 45.83
- Graph of Agents: Principled Long Context Modeling by Emergent Multi-Agent Collaboration
GoA beats RAG · Average F1 [Qwen3 8B, 8K context]
48.98 vs 47.29
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · HotpotQA [Qwen2.5-7B-Inst]
67.3 vs 53.8
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · Narr.QA [Qwen2.5-7B-Inst]
20.7 vs 14.2
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · CR [Qwen2.5-7B-Inst]
15.1 vs 7.54
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · MIR [Qwen2.5-7B-Inst]
12.6 vs 5.29
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · Code.Debug [Qwen2.5-7B-Inst]
24.8 vs 22.1
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · Code.Run [Qwen2.5-7B-Inst]
21.2 vs 3.2
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · GraphWalks [Qwen2.5-7B-Inst]
29.8 vs 25.6
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · MRCR [Qwen2.5-7B-Inst]
16.3 vs 8.8
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats RAG · HotpotQA [Llama3.1-7B-Inst]
69.4 vs 55.1
Newer alternatives
Recent methods in the same sub-problem, not yet superseded in the knowledge base.