List your company

Multiple-choice Reasoning on GPQA (test)

65.7AccuracyRouteGoT
40.42846.98953.5560.111Mar 6, 2026

Evaluation results

MethodLinks
RouteGoTBackbone={Qwen3-4B, 8B, 30B}2026.0365.73,352
AGoTBackbone=Qwen3-30B2026.0364.612,179
CoTBackbone=Qwen3-30B2026.0363.14,965
KNNBackbone={Qwen3-4B, 8B, 30B}2026.0361.13,292
RandomBackbone={Qwen3-4B, 8B, 30B}2026.0360.15,658
GoT*Backbone=Qwen3-30B2026.0359.69,468
EmbedLLMBackbone={Qwen3-4B, 8B, 30B}2026.0359.611,369
RouteLLMBackbone={Qwen3-4B, 8B, 30B}2026.0357.63,640
RTRBackbone={Qwen3-4B, 8B, 30B}2026.0356.63,224
ToTBackbone=Qwen3-30B2026.0344.99,077
IOBackbone=Qwen3-30B2026.0341.47