reasoning benchmark
MASK
MASK is a collection of 1000 questions measuring whether models faithfully report their beliefs when pressured to lie. It operationalizes deception as the rate at which the model lies, i.e., knowingly making false statements intended to be received as true. Lower dishonesty rates indicate better honesty.
Updated Aug 11, 2026
Models1
Model coverage1
MetricScore
EvidenceB
MASK Ranking
Higher score ranks better on this benchmark.
| 1 | XA | 51.0% | 100.0% | 1 | C |
MASK Highlights
The leading models and scores on this benchmark.
What is MASK?
What MASK measures and how its scores work.
MASK is a collection of 1000 questions measuring whether models faithfully report their beliefs when pressured to lie. It operationalizes deception as the rate at which the model lies, i.e., knowingly making false statements intended to be received as true. Lower dishonesty rates indicate better honesty.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- MASK
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- mask|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about MASK.
Which model scores highest on MASK?
Grok-4.1 Thinking is currently ranked first with 51.0%.
What does MASK measure?
MASK is a collection of 1000 questions measuring whether models faithfully report their beliefs when pressured to lie. It operationalizes deception as the rate at which the model lies, i.e., knowingly making false statements intended to be received as true. Lower dishonesty rates indicate better honesty.
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.