safety benchmark
XSTest
XSTest is a test suite designed to identify exaggerated safety behaviours in large language models. It comprises 450 prompts: 250 safe prompts across ten prompt types that well-calibrated models should not refuse to comply with, and 200 unsafe prompts as contrasts that models should refuse. The benchmark systematically evaluates whether models refuse to respond to clearly safe prompts due to overly cautious safety mechanisms.
Updated Aug 11, 2026
Models3
Model coverage3
MetricScore
EvidenceB
XSTest Ranking
Higher score ranks better on this benchmark.
| 1 | GO | 98.8% | 100.0% | 3 | C | |
| 2 | GO | 97.0% | 50.0% | 3 | C | |
| 3 | GO | 92.6% | 0.0% | 3 | C |
XSTest Score Distribution
A closer view of the leading scores on this benchmark.
XSTest Highlights
The leading models and scores on this benchmark.
What is XSTest?
What XSTest measures and how its scores work.
XSTest is a test suite designed to identify exaggerated safety behaviours in large language models. It comprises 450 prompts: 250 safe prompts across ten prompt types that well-calibrated models should not refuse to comply with, and 200 unsafe prompts as contrasts that models should refuse. The benchmark systematically evaluates whether models refuse to respond to clearly safe prompts due to overly cautious safety mechanisms.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- XSTest
- Modality
- text
- Primary category
- safety
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- xstest|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about XSTest.
Which model scores highest on XSTest?
Gemini 1.5 Pro is currently ranked first with 98.8%.
What does XSTest measure?
XSTest is a test suite designed to identify exaggerated safety behaviours in large language models. It comprises 450 prompts: 250 safe prompts across ten prompt types that well-calibrated models should not refuse to comply with, and 200 unsafe prompts as contrasts that models should refuse. The benchmark systematically evaluates whether models refuse to respond to clearly safe prompts due to overly cautious safety mechanisms.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
3 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.