Compare memory systems
21 head-to-head pages, one per pair, all from the same 30-question pilot of LongMemEval-S (cleaned, 2025-09). Each one-line verdict calls a pair "statistically tied" when the two 95% intervals overlap: at this sample size, most memory systems cannot yet be separated on accuracy.
Cognee comparisons
Cognee results- Cognee vs Full contextStatistically tied on accuracy at n=30; Full context 11× cheaper
- Cognee vs GraphitiGraphiti: pilot result withdrawn, rerun in progress
- Cognee vs HindsightTied on accuracy (90.0% each); Cognee 1.2× cheaper
- Cognee vs LangMemCognee more accurate; LangMem 2.3× cheaper
- Cognee vs Mem0Statistically tied on accuracy at n=30; Mem0 1.5× cheaper
- Cognee vs Plain RAGStatistically tied on accuracy at n=30; Plain RAG 26× cheaper
Hindsight comparisons
Hindsight results- Hindsight vs CogneeTied on accuracy (90.0% each); Cognee 1.2× cheaper
- Hindsight vs Full contextStatistically tied on accuracy at n=30; Full context 13× cheaper
- Hindsight vs GraphitiGraphiti: pilot result withdrawn, rerun in progress
- Hindsight vs LangMemHindsight more accurate; LangMem 2.8× cheaper
- Hindsight vs Mem0Statistically tied on accuracy at n=30; Mem0 1.8× cheaper
- Hindsight vs Plain RAGStatistically tied on accuracy at n=30; Plain RAG 31× cheaper
Mem0 comparisons
Mem0 results- Mem0 vs CogneeStatistically tied on accuracy at n=30; Mem0 1.5× cheaper
- Mem0 vs Full contextTied on accuracy (83.3% each); Full context 7.1× cheaper
- Mem0 vs GraphitiGraphiti: pilot result withdrawn, rerun in progress
- Mem0 vs HindsightStatistically tied on accuracy at n=30; Mem0 1.8× cheaper
- Mem0 vs LangMemMem0 more accurate; LangMem 1.5× cheaper
- Mem0 vs Plain RAGStatistically tied on accuracy at n=30; Plain RAG 17× cheaper
LangMem comparisons
LangMem results- LangMem vs CogneeCognee more accurate; LangMem 2.3× cheaper
- LangMem vs Full contextFull context more accurate and 4.7× cheaper
- LangMem vs GraphitiGraphiti: pilot result withdrawn, rerun in progress
- LangMem vs HindsightHindsight more accurate; LangMem 2.8× cheaper
- LangMem vs Mem0Mem0 more accurate; LangMem 1.5× cheaper
- LangMem vs Plain RAGPlain RAG more accurate and 11× cheaper
Graphiti comparisons
Graphiti results- Graphiti vs CogneeGraphiti: pilot result withdrawn, rerun in progress
- Graphiti vs Full contextGraphiti: pilot result withdrawn, rerun in progress
- Graphiti vs HindsightGraphiti: pilot result withdrawn, rerun in progress
- Graphiti vs LangMemGraphiti: pilot result withdrawn, rerun in progress
- Graphiti vs Mem0Graphiti: pilot result withdrawn, rerun in progress
- Graphiti vs Plain RAGGraphiti: pilot result withdrawn, rerun in progress
Plain RAG comparisonsbaseline
Plain RAG results- Plain RAG vs CogneeStatistically tied on accuracy at n=30; Plain RAG 26× cheaper
- Plain RAG vs Full contextStatistically tied on accuracy at n=30; Plain RAG 2.4× cheaper
- Plain RAG vs GraphitiGraphiti: pilot result withdrawn, rerun in progress
- Plain RAG vs HindsightStatistically tied on accuracy at n=30; Plain RAG 31× cheaper
- Plain RAG vs LangMemPlain RAG more accurate and 11× cheaper
- Plain RAG vs Mem0Statistically tied on accuracy at n=30; Plain RAG 17× cheaper
Full context comparisonsbaseline
Full context results- Full context vs CogneeStatistically tied on accuracy at n=30; Full context 11× cheaper
- Full context vs GraphitiGraphiti: pilot result withdrawn, rerun in progress
- Full context vs HindsightStatistically tied on accuracy at n=30; Full context 13× cheaper
- Full context vs LangMemFull context more accurate and 4.7× cheaper
- Full context vs Mem0Tied on accuracy (83.3% each); Full context 7.1× cheaper
- Full context vs Plain RAGStatistically tied on accuracy at n=30; Plain RAG 2.4× cheaper