reasoning benchmark
OpenBookQA
OpenBookQA is a question-answering dataset modeled after open book exams for assessing human understanding. It contains 5,957 multiple-choice elementary-level science questions that probe understanding of 1,326 core science facts and their application to novel situations, requiring combination of open book facts with broad common knowledge through multi-hop reasoning.
Updated Aug 7, 2026
Published models5
Registry coverage5
MetricScore
EvidenceB
OpenBookQA leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | MI | 89.6% | 100.0% | 5 | C | |
| 2 | MI | 79.2% | 75.0% | 5 | C | |
| 3 | MI | 79.2% | 50.0% | 5 | C | |
| 4 | MA | 60.6% | 25.0% | 5 | C | |
| 5 | NR | 49.4% | 0.0% | 5 | C |
Score distribution
Top published rows on the benchmark's original scale.
OpenBookQA highlights
The top published results on this benchmark's own scale.
What is OpenBookQA?
Definition and scoring fields from the benchmark registry.
OpenBookQA is a question-answering dataset modeled after open book exams for assessing human understanding. It contains 5,957 multiple-choice elementary-level science questions that probe understanding of 1,326 core science facts and their application to novel situations, requiring combination of open book facts with broad common knowledge through multi-hop reasoning.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- OpenBookQA
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- openbookqa|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about OpenBookQA.
Which model scores highest on OpenBookQA?
Phi-3.5-MoE-instruct is currently ranked first with 89.6%.
What does OpenBookQA measure?
OpenBookQA is a question-answering dataset modeled after open book exams for assessing human understanding. It contains 5,957 multiple-choice elementary-level science questions that probe understanding of 1,326 core science facts and their application to novel situations, requiring combination of open book facts with broad common knowledge through multi-hop reasoning.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
5 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.