reasoning benchmark
FlenQA
Flexible Length Question Answering dataset for evaluating the impact of input length on reasoning performance of language models, featuring True/False questions embedded in contexts of varying lengths (250-3000 tokens) across three reasoning tasks: Monotone Relations, People In Rooms, and simplified Ruletaker
Updated Aug 7, 2026
Published models2
Registry coverage2
MetricScore
EvidenceB
FlenQA leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | MI | 97.9% | 100.0% | 2 | C | |
| 2 | MI | 97.7% | 0.0% | 2 | C |
Score distribution
Top published rows on the benchmark's original scale.
FlenQA highlights
The top published results on this benchmark's own scale.
What is FlenQA?
Definition and scoring fields from the benchmark registry.
Flexible Length Question Answering dataset for evaluating the impact of input length on reasoning performance of language models, featuring True/False questions embedded in contexts of varying lengths (250-3000 tokens) across three reasoning tasks: Monotone Relations, People In Rooms, and simplified Ruletaker
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- FlenQA
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- flenqa|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about FlenQA.
Which model scores highest on FlenQA?
Phi 4 Reasoning Plus is currently ranked first with 97.9%.
What does FlenQA measure?
Flexible Length Question Answering dataset for evaluating the impact of input length on reasoning performance of language models, featuring True/False questions embedded in contexts of varying lengths (250-3000 tokens) across three reasoning tasks: Monotone Relations, People In Rooms, and simplified Ruletaker
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
2 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.