List your company

DeepSearch Question Answering on Humanity's Last Exam

14.7NS ScoreAPPO
2.225.468.711.94Jun 10, 2026

Evaluation results

MethodLinks
APPOBackbone=Qwen3-8B, Method Strategy=APPO, Method Category=RL-based Method2026.0614.78.31313.4
Qwen3-32B-thinkingBackbone=Qwen3-32B, Method Category=Reference2026.0614.69.88.412.6
o1-previewBackbone=o1-preview, Method Category=Reference2026.0612.98.16.611.1
QwQ-32BBackbone=QwQ-32B, Method Category=Reference2026.0611.57.35.29.6
APPOBackbone=Qwen3-14B, Method Strategy=APPO, Method Category=RL-based Method2026.0611.512.814.512.3
ARPOBackbone=Qwen3-14B, Method Strategy=ARPO, Method Category=RL-based Method2026.0610.310.713.710
DeepSeek-R1-671BBackbone=DeepSeek-R1-671B, Method Category=Reference2026.068.58.19.38.6
GRPOBackbone=Qwen3-8B, Method Strategy=GRPO, Method Category=RL-based Method2026.067.9410.57.8
GRPOBackbone=Qwen3-14B, Method Strategy=GRPO, Method Category=RL-based Method2026.067.96.712.68.6
Search-o1Backbone=Qwen3-8B, Method Strategy=Search-o1, Method Category=Single-Enhanced Method2026.067.62.75.36.4
WebThinkerBackbone=Qwen3-8B, Method Strategy=WebThinker, Method Category=Single-Enhanced Method2026.067.346.36.6
ARPOBackbone=Qwen3-8B, Method Strategy=ARPO, Method Category=RL-based Method2026.067.36.715.88.8
WebThinkerBackbone=Qwen3-14B, Method Strategy=WebThinker, Method Category=Single-Enhanced Method2026.06749.57
DeepSeek-R1-32BBackbone=DeepSeek-R1-32B, Method Category=Reference2026.066.65.16.56.4
Search-o1Backbone=Qwen3-14B, Method Strategy=Search-o1, Method Category=Single-Enhanced Method2026.066.4410.56.8
ReActBackbone=Qwen3-14B, Method Strategy=ReAct, Method Category=Single-Enhanced Method2026.065.85.310.56.6
Vanilla RAGBackbone=Qwen3-14B, Method Strategy=Vanilla RAG, Method Category=Single-Enhanced Method2026.065.56.39.46
Qwen3-14BBackbone=Qwen3-14B, Method Strategy=Base, Method Category=RL-based Method2026.065.56.711.66.8
Vanilla RAGBackbone=Qwen3-8B, Method Strategy=Vanilla RAG, Method Category=Single-Enhanced Method2026.065.11.612.95.8
ReActBackbone=Qwen3-8B, Method Strategy=ReAct, Method Category=Single-Enhanced Method2026.064.246.34.6
Qwen3-8BBackbone=Qwen3-8B, Method Strategy=Base, Method Category=RL-based Method2026.063.92.78.44.6
GPT-4oBackbone=GPT-4o, Method Category=Reference2026.062.71.23.22.6