Streaming Long-context Question Answering on STEM-Bench Friends benchmark leaderboard · SOTA2 Research

Streaming Long-context Question Answering on STEM-Bench Friends

0.318TimeProStream (-w/o Mtree)

Evaluation Results

Method	Links
ProStream (-w/o Mtree)2026.03		0.318	0.097	0.304	0.703	0.48	0.581	0.466
ProStream (-w/o Mpend)2026.03		0.32	0.09	0.356	0.704	0.482	0.58	0.473
ProStream2026.03		0.373	0.099	0.36	0.801	0.491	0.588	0.476
ProStream (-w/o B)2026.03		0.469	0.034	0.149	0.75	0.276	0.364	0.202
RAPTOR2026.03		0.513	0.035	0.157	0.746	0.396	0.367	0.214
RAG2026.03		0.923	0.061	0.231	0.768	0.345	0.502	0.309
GraphRAG2026.03		0.95	0.055	0.227	0.775	0.46	0.465	0.309
A-Mem2026.03		1.29	0.053	0.2	0.742	0.386	0.376	0.266
Summary2026.03		2.352	0.042	0.18	0.762	0.334	0.385	0.246
MemoRAG2026.03		4.146	0.043	0.178	0.771	0.575	0.355	0.242
Full Context2026.03		4.87	0.057	0.241	0.732	0.36	0.706	0.505
RQ-RAG2026.03		5.098	0.043	0.184	0.758	0.372	0.439	0.313
HippoRAG22026.03		6.539	0.041	0.165	0.742	0.529	0.479	0.249
MemGAS2026.03		7.982	0.012	0.072	0.53	0.433	0.749	0.466