Humanity's Last Exam

Last updated

Humanity's Last Exam (HLE) is a language model benchmark consisting of 2,500 questions across a broad range of subjects. It was created jointly by the Center for AI Safety and Scale AI.

Contents

Creation

Stanford HAI's AI Index 2025 Annual Report cites Humanity's Last Exam as one of the "more challenging benchmarks" developed in response to the popular AI benchmarks having reached "saturation". [1] The test has been described as the brainchild of Dan Hendrycks, a machine learning researcher and the director of the Center for AI Safety, who stated that he was inspired to create the test after a conversation with Elon Musk, who thought the existing language model benchmarks, such as the MMLU, were too easy. Hendrycks worked with Scale AI to compile the questions. [2] The questions were crowdsourced from subject matter experts from various institutions across the world. [3] [4] The questions were first filtered by the leading AI models; if the models failed to answer the question or did worse than random guessing on the multiple-choice questions, they were reviewed by human experts in two rounds and approved for inclusion in the dataset. The submitters of the top-rated questions were given prize money from a pool of 500,000 U.S. dollars—$5000 for each of the top 50 questions and $500 for the next 500. After the initial release, a "community feedback bug bounty program" was opened to "identify and remove major errors in the dataset". [4]

Composition

The benchmark consists of 2,500 questions in the publicly released set. The paper classifies the questions into the following broad subjects: mathematics (41%), physics (9%), biology/medicine (11%), humanities/social science (9%), computer science/artificial intelligence (10%), engineering (4%), chemistry (7%), and other (9%). Around 14% of the questions require the ability to understand both text and images, i.e., multi-modality. 24% of the questions are multiple-choice; the rest are short-answer, exact-match questions. A private set is also maintained to test for benchmark overfitting. [4]

An example question: [2]

Hummingbirds within Apodiformes uniquely have a bilaterally paired oval bone, a sesamoid embedded in the caudolateral portion of the expanded, cruciate aponeurosis of insertion of m. depressor caudae. How many paired tendons are supported by this sesamoid bone? Answer with a number.

An independent investigation by FutureHouse, published in July 2025, [5] suggests that around 30% of the HLE answers for text-only chemistry and biology questions may be incorrect.

Results

Performance of various models on the benchmark
OrganizationModelAccuracy (%) ↑Calibration Error (%) ↓
OpenAI GPT-5 25.3250
Google DeepMind Gemini 2.5 Pro Preview (06-05) 21.6472
Anthropic Claude Opus 4.1 (Thinking) 11.5271
Zhipu AI GLM 4.58.3279
Meta AI Llama 4 Maverick 5.6883
Mistral AI Mistral Medium 3 4.5277
Amazon Web Services Nova Pro4.4080
Source: Scale AI. 15 August 2025.
Performance of various non-multimodal models on the text-only subset of the benchmark
OrganizationModelAccuracy (%) ↑Calibration Error (%) ↓
Alibaba Cloud Qwen3-235B-A22B-Thinking-2507 15.4378
DeepSeek DeepSeek-R1-0528 14.0478
OpenAI o3-mini (high) 13.3780
Moonshot AI Kimi-K2-Instruct 4.6882
Amazon Web Services Nova Micro4.4184
Source: Scale AI. 5 August 2025.

References

  1. Maslej, Nestor; et al. (April 2025). The AI Index 2025 Annual Report (PDF) (Report). Institute for Human-Centered AI. pp. 141–142.
  2. 1 2 Roose, Kevin (23 January 2025). "When A.I. Passes This Test, Look Out". New York Times. Archived from the original on 29 January 2025. Retrieved 24 January 2025.
  3. Dastin, Jeffrey; Paul, Katie (16 September 2024). "AI experts ready 'Humanity's Last Exam' to stump powerful tech". Reuters. Archived from the original on 8 April 2025. Retrieved 24 January 2025.
  4. 1 2 3 Phan, Long; et al. (2025). "Humanity's Last Exam". arXiv: 2501.14249 [cs.LG].
  5. "About 30% of Humanity's Last Exam chemistry/biology answers are likely wrong | FutureHouse". www.futurehouse.org. Retrieved 2025-07-26.