List your company

Question Answering on GAIA (held-out)

47.7EMWebExpert+SFT
12.23621.44330.6539.857Feb 3, 2026

Evaluation results

MethodLinks
WebExpert+SFTBackbone=QwQ-32B, Training=SFT2026.0247.7
WebExpertBackbone=QwQ-32B2026.0246.2
WebThinker-32B-BaseBackbone=32B-Base2026.0244.7
Search-o1-32BBackbone=o1-32B2026.0239.8
RAG-QwQ-32BStrategy=RAG2026.0232
QwQ-32BStrategy=Direct2026.0213.6