GPT-4.0 and GPT-5.0 Performance on the 2024 AAP Periodontology In-Service Exam

Descriptive performance data for two large language models evaluated on a professional medical examination. The dataset contains results for GPT-4.0 and GPT-5.0 on the 2024 American Academy of Periodontology In-Service Examination, comprising 331 multiple-choice questions. Results are presented as the number and percentage of correct responses for each exam section under two testing conditions, along with completion rates and missing responses.

Use Cases

• Benchmarking LLM accuracy on domain-specific knowledge based on the 331-question periodontology exam.
• Comparing model performance under different testing conditions based on the Full Test (FT) versus Individual Question (IQ) presentation modes.
• Analyzing performance variation across subject domains within a professional medical field based on the reported exam sections.

Strengths

• Results are based on a standardized, professionally administered examination containing 331 multiple-choice questions.
• Performance is reported for two distinct LLM models (GPT-4.0 and GPT-5.0) under two different testing conditions.
• Data includes completion rates and missing responses, providing insight into model reliability.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small at 5.5 KB, indicating limited scope.

Provenance

Source

Prita Abhay Dhaimade

Collection Method

Evaluation of GPT-4.0 and GPT-5.0 using the 2024 AAP In-Service Examination.

Time Range

2024

Freshness

Last updated 2026-05-29 17:32:26; freshness should be verified.