List your company

Long-term memory assessment on LONGMEMEVAL

86.46LLM-Judge Multi-sessionMRAgent*
25.93241.64657.3673.074Jun 4, 2026

Evaluation results

MethodLinks
MRAgent*Retrieval Model=Claude, Memory Construction Model=Gemini2026.0686.4692.8585.7178.5786.76
MRAgentBackbone=Gemini2026.0668.4292.8568.4266.6772.95
MRAgent*Retrieval Backbone=Claude, Memory Construction Backbone=Gemini2026.0666.3182.4160.115.5886.76
MemoryOSBackbone=Gemini2026.0656.3987.1438.3546.6754.92
RAGBackbone=Gemini2026.0654.8985.7142.8633.3354.65
LangMemBackbone=Gemini2026.0652.6378.5745.7136.6753.77
Mem0Backbone=Gemini2026.0650.3878.5745.114053.01
MRAgentBackbone=Gemini2026.0649.9280.9950.1623.9672.95
MemoryOSBackbone=Gemini2026.0645.0279.535.679.5354.92
RAGBackbone=Gemini2026.064577.9443.885.4254.65
A-MemBackbone=Gemini2026.0642.859045.1146.4352.98
LangMemBackbone=Gemini2026.0641.1472.4336.86.3253.77
Mem0Backbone=Gemini2026.0637.5372.4335.645.8953.01
A-MemBackbone=Gemini2026.0628.2675.334.189.0952.98