Unverified50% confidenceFactExact time
GPQA是由纽约大学等机构开发的博士级科学推理基准，题目由领域博士生编写并经过交叉验证
1
Sources
50%
Confidence
Long-term
Relevance
7/9/2026
First Seen
Sources
OpenAI's Head of Evals: We've Been Consistently Underestimating What AI Models Can Really Do
bilibiliGelai_AI7/8/2026
Related Claims
UnverifiedGPQA was designed by researchers at New York University, covers graduate-level knowledge in biology, chemistry, physics, and other fields, and human experts average approximately 65%.78% similarUnverifiedThe information was disclosed by a researcher from Princeton University's AI lab.52% similarUnverifiedCo-Scientist内部包含专门负责文献检索、假设生成、批判性评估和实验设计的子智能体51% similarUnverifiedNVIDIA's developer blog published a technical article on how to add a Deep Research Skill to Agent frameworks.50% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/362869API
curl https://kongchang.com/api/v1/knowledge/claims/362869MCP
get_claim(id=362869)