Is RAPTOR superseded?

RAPTOR (Retrieval-augmented generation): heavily superseded — a standard baseline that newer methods routinely beat. 7 paper(s) critique it, 21 beat it on benchmarks — #5 of 1179 most-superseded. Sub-problem: cluster led by RAG. Newer alternatives in the same sub-problem include Narrative Knowledge Weaver, IA-RAG, MemGraphRAG, LegalGraphRAG, R2C.

Method Drift›Retrieval-augmented generation

Heavily superseded#5 of 1,179 most-superseded

RAPTOR

RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval

Retrieval-augmented generation · first seen Jan 31, 2024

heavily superseded — a standard baseline that newer methods routinely beat

7 papers critique it · 21 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites RAPTOR as a baseline.

“its exclusive focus on semantic similarity presents two key limitations: (1) it often misses important connected relationships that lack direct semantic similarity, hindering multi-hop reasoning capabilities, and (2) its rigid hierarchical structure can fragment interconnected knowledge”
— MacRAG: Compress, Slice, and Scale-up for Multi-Scale Adaptive Context RAG
“Despite their effectiveness in entity-rich tasks, KG-based methods face scalability and adaptability challenges and often require substantial resources on the corpus processing side.”
— Efficient Dynamic Clustering-Based Document Compression for Retrieval-Augmented-Generation
“these groupings are fixed at ingestion and do not adapt to the specific structural or layout demands of individual queries.”
— LAD-RAG: Layout-aware Dynamic RAG for Visually-Rich Document Understanding
“As shown in fig:corpusindex(b), RAPTOR's retrieval accuracy drops significantly when the search space expands to corpus-level with millions of tokens.”
— Hierarchical Abstract Tree for Cross-Document Retrieval-Augmented Generation
“RAPTOR's performance deteriorates substantially on the simple and multi-hop QA tasks due to the noise introduced into the retrieval corpora by its LLM summarization mechanism.”
— From RAG to Memory: Non-Parametric Continual Learning for Large Language Models
“While existing approaches offline-encode hierarchical information into fixed representations (e.g., summaries or embeddings), our framework online-perceives document structure through dynamic routing.”
— Equipping Retrieval-Augmented Large Language Models with Document Structure Awareness
“Existing methods merely focus on the surface form of the utterance, relying on semantic similarity through global dense retrieval over stored memory traces~sarthi2024raptor or traversal heuristics over the predefined structures~xu2025amem,jiang2026magma.”
— Goal-Oriented Reasoning for RAG-based Memory in Conversational Agentic LLM Systems

Beaten on benchmarks

Head-to-head results where a newer method reports beating RAPTOR. Values are copied from the source paper's tables — verify against the cited paper.

OG-RAG beats RAPTOR · C-Recall [Llama-3-8B]
0.82 vs 0.56
OG-RAG: Ontology-Grounded Retrieval-Augmented Generation For Large Language Models
OG-RAG beats RAPTOR · A-Corr [Llama-3-8B]
0.40 vs 0.34
OG-RAG: Ontology-Grounded Retrieval-Augmented Generation For Large Language Models
OG-RAG beats RAPTOR · C-Recall [Llama-3-70B]
0.84 vs 0.55
OG-RAG: Ontology-Grounded Retrieval-Augmented Generation For Large Language Models
OG-RAG beats RAPTOR · A-Corr [Llama-3-70B]
0.54 vs 0.41
OG-RAG: Ontology-Grounded Retrieval-Augmented Generation For Large Language Models
OG-RAG beats RAPTOR · C-Recall [GPT-4o]
0.84 vs 0.54
OG-RAG: Ontology-Grounded Retrieval-Augmented Generation For Large Language Models
OG-RAG beats RAPTOR · A-Corr [GPT-4o]
0.48 vs 0.34
OG-RAG: Ontology-Grounded Retrieval-Augmented Generation For Large Language Models
OG-RAG beats RAPTOR · A-Corr [GPT-4o-mini]
0.50 vs 0.42
OG-RAG: Ontology-Grounded Retrieval-Augmented Generation For Large Language Models
LegalGraphRAG beats RAPTOR · Accuracy All [Qwen3-8B]
41.3 vs 34.0
LegalGraphRAG: Multi-Agent Graph Retrieval-Augmented Generation for Reliable Legal Reasoning
LegalGraphRAG beats RAPTOR · Accuracy All [DeepSeek-V3.1]
49.9 vs 44.4
LegalGraphRAG: Multi-Agent Graph Retrieval-Augmented Generation for Reliable Legal Reasoning
LegalGraphRAG beats RAPTOR · Accuracy All [GPT-4o-mini]
40.9 vs 30.5
LegalGraphRAG: Multi-Agent Graph Retrieval-Augmented Generation for Reliable Legal Reasoning
Corpus2Skill beats RAPTOR · F1 [WixQA benchmark]
0.456 vs 0.398
Don't Retrieve, Navigate: Distilling Enterprise Knowledge into Navigable Agent Skills for QA and RAG
Corpus2Skill beats RAPTOR · BERTScore-F1 [WixQA benchmark]
0.862 vs 0.839
Don't Retrieve, Navigate: Distilling Enterprise Knowledge into Navigable Agent Skills for QA and RAG

Newer alternatives

Recent methods in the same sub-problem, not yet superseded in the knowledge base.