Long-context Memory Management on LoCoMo (Precision, Recall, F1, Occupancy, Reward) benchmark leaderboard · SOTA2 Research

Long-context Memory Management on LoCoMo (Precision, Recall, F1, Occupancy, Reward)

31.3PrecisionOSL-MR

Evaluation Results

Method	Links
OSL-MR2026.06		31.3	35.87	29.52	73.67	252.7
OSL-MR2026.06		27.41	53.2	30.23	76.38	305.16
BC-learner2026.06		23.03	44.41	25.56	78.97	260.91
BC-learner2026.06		22.3	24.06	20.61	76.32	196.83
OSL-MR2026.06		11.8	10.83	10.15	66.83	145.56
BC-learner2026.06		9.68	8.26	8.11	65.67	137.06
Mixed-Score2026.06		3.91	33.75	6.85	99.56	132.48
Mixed-Score2026.06		3.25	15.55	5.22	99.2	112.4
Mixed-Score2026.06		2.43	6.99	3.52	97.89	110.79
GA-heuristic2026.06		0.41	5.08	0.74	98.98	9.99
GA-heuristic2026.06		0.38	3.02	0.68	98.08	67.68
GA-LLM2026.06		0.3	3.98	0.56	99.05	6.4
GA-heuristic2026.06		0.26	1	0.41	96.82	95.04
GA-LLM2026.06		0.22	1.23	0.36	98.35	63.68
GA-LLM2026.06		0.17	0.62	0.26	96.37	94.41
Recency2026.06		0.12	1.45	0.21	99.19	0
Recency2026.06		0.1	0.62	0.17	98.9	61.95
Recency2026.06		0	0	0	98.06	92.44