Share your thoughts, 1 month free Claude Pro on usSee more

Graph Reasoning on GraphArena

100Average Acc

AgentSquare

Updated 22d ago

Evaluation Results

Method	Links
AgentSquare 2026.05		100	-	-
MA-GTS 2026.05		100	-	-
w/o SCA Routing 2026.05		99.5	-	-
w/o Tool Growth 2026.05		99.5	-	-
w/o Protocol-Reliability Obj 2026.05		99.5	-	-
EGL-SCA (Full) 2026.05		99.5	-	-
w/o Instruction Evol. 2026.05		91	-	-
FixedSolver 2026.05		67	-	-
ReAct 2026.05		65.5	-	-
NPG-Muse-8B 2025.08		54.4	63.2	32.9
GEPA 2026.05		47.5	-	-
Chain-of-Thought (CoT) 2026.05		42	-	-
Qwen2.5-14B-Instruct-1M 2025.08		40.9	48.1	15.8
NPG-Muse-7B 2025.08		36.2	47.7	10.7
w/o Tool Use 2026.05		36	-	-
Qwen2.5-7B-Ins-1M 2025.08		32.9	37.2	15.2
Qwen3-14B-Base 2025.08		32.6	31.6	14.8
Qwen3-8B-Base 2025.08		25.2	20.8	9.5
ExpeL 2026.05		18	-	-
Direct Prompting 2026.05		16	-	-
Few-Shot 2026.05		8	-	-
Reflexion 2026.05		4.5	-	-