Gemini 3.0 Pro scores 68% on HLE, dubbed 'Humanity's Last Exam'
Original title:From the original HLE paper:
AI summary
Dan Hendrycks reposted a discussion on X about the HLE (Humanity's Last Exam) benchmark, saying reports indicate Gemini 3.0 Pro scored 68% on it. The accompanying discussion notes that current LLMs still perform poorly on HLE, but recent benchmarks have saturated quickly; if a model exceeds 50% accuracy by the end of 2025, it would demonstrate expert-level closed-book, verifiable research and reasoning ability, and HLE is seen as the final academic benchmark for AI.
Key Takeaways
- Dan Hendrycks reposted a discussion on X about the HLE (Humanity's Last Exam) benchmark,saying…
- 0 Pro scored 68% on it.
- The accompanying discussion notes that current LLMs still perform poorly on HLE,but recent…
- if a model exceeds 50% accuracy by the end of 2025,it would demonstrate expert-level…
Quick Facts
- Published
- Updated
- Source
- X:Dan Hendrycks (@hendrycks)
- Category
- Papers & Research
- AI score
- 80
HLE 是一个聚焦技术知识与推理能力的高难度学术基准,其设计意图是衡量模型是否具备真正‘科学智能’而非仅靠模式匹配。该消息首次公开披露主流大模型在该基准上的实测分数,且由领域权威学者发布,具有较高可信度和参考价值,对判断当前 AI 真实能力边界至关重要。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Gemini 3.0 Pro scores 68% on HLE, dubbed 'Humanity's Last Exam'. November 5, 2025. https://new.nat6.net/en/news/ig04jhdb9trrwn7arykt83gqg/