List your company

Scientific Question Answering on GPQA Diamond (OOD) (test)

89.84AccuracySkillFlow
60.595268.187675.7883.3724May 13, 2026

Evaluation results

MethodLinks
SkillFlowMethod Type=Ours, Backbone Model=Qwen3.5-9B2026.0589.84
FlowSteerMethod Type=Agent+RL, Backbone Model=Qwen3.5-9B2026.0584.38
AgentFlowMethod Type=Agent+RL, Backbone Model=Qwen3.5-9B2026.0581.25
SkillRLMethod Type=Agent+RL, Backbone Model=Qwen3.5-9B2026.0578.13
AFlowMethod Type=Search-based workflow, Backbone Model=Qwen3.5-9B2026.0575.78
Qwen3.5Method Type=GRPO, Backbone Model=Qwen3.5-9B2026.0575
v4-flashMethod Type=Baseline, Backbone Model=v4-flash2026.0573.44
Qwen3.5Method Type=SFT, Backbone Model=Qwen3.5-9B2026.0571.88
Qwen3.5Method Type=Baseline, Backbone Model=Qwen3.5-9B2026.0561.72