SimpleQA what it measures, how it is scored and who leads it

Name: SimpleQA
Creator: OpenAI
Published: 2024-11-07

Knowledge · OpenAI · introduced Nov 7, 2024 · Correct (and F-score)

About this benchmark

Task: Answer short fact-seeking questions with a single correct answer, or decline.
Dataset: Short questions collected adversarially against GPT-4 answers.
Method: Each answer graded correct, incorrect or not attempted by a model grader against the reference.
Metric: Correct (and F-score)
Organization: OpenAI
Introduced: Nov 7, 2024
Paper: Measuring short-form factuality in large language models (arXiv 2411.04368)
Home: github.com/openai/simple-evals

Newest first. New versions are added, never rewritten.

Version	Date	What changed
1Released with the paper.	Nov 7, 20242 years ago	Released with the paper.

Sources: each benchmark's paper (arXiv) and its official site or leaderboard. Scores are the boards captured here every day, exactly as published.