factuality benchmark
LongFact
LongFact evaluates factual precision over long-form generations containing many individual claims. Each claim is extracted and verified, and the model is scored on claim-level precision, measuring whether extended responses introduce unsupported or false statements.
Updated Aug 11, 2026
Models1
Model coverage1
MetricScore
EvidenceB
LongFact Ranking
Higher score ranks better on this benchmark.
| 01 | MI | 98.0% | 100.0% | 1 | C |
LongFact Highlights
The leading models and scores on this benchmark.
What is LongFact?
What LongFact measures and how its scores work.
LongFact evaluates factual precision over long-form generations containing many individual claims. Each claim is extracted and verified, and the model is scored on claim-level precision, measuring whether extended responses introduce unsupported or false statements.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- LongFact
- Modality
- text
- Primary category
- factuality
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- longfact|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about LongFact.
Which model scores highest on LongFact?
MAI-Thinking-1 is currently ranked first with 98.0%.
What does LongFact measure?
LongFact evaluates factual precision over long-form generations containing many individual claims. Each claim is extracted and verified, and the model is scored on claim-level precision, measuring whether extended responses introduce unsupported or false statements.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 model results are currently shown.
Does this benchmark affect the overall score?
No. This benchmark is shown for reference but does not contribute to the overall score.