safety benchmark
CyberSecEval 4
CyberSecEval 4 is an evaluation suite covering cybersecurity-related capabilities and risks of large language models. The insecure-code-generation tracks measure whether a model produces vulnerable code: the Instruct track presents coding requests designed to elicit known insecure patterns, while the Autocomplete track prompts the model with code context leading up to a known insecure pattern, with vulnerabilities detected via static analysis.
Updated Aug 7, 2026
Published models1
Registry coverage1
MetricScore
EvidenceB
CyberSecEval 4 leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | MI | 63.0% | 100.0% | 1 | C |
CyberSecEval 4 highlights
The top published results on this benchmark's own scale.
What is CyberSecEval 4?
Definition and scoring fields from the benchmark registry.
CyberSecEval 4 is an evaluation suite covering cybersecurity-related capabilities and risks of large language models. The insecure-code-generation tracks measure whether a model produces vulnerable code: the Instruct track presents coding requests designed to elicit known insecure patterns, while the Autocomplete track prompts the model with code context leading up to a known insecure pattern, with vulnerabilities detected via static analysis.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- CyberSecEval 4
- Modality
- text
- Primary category
- safety
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- cyberseceval-4|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about CyberSecEval 4.
Which model scores highest on CyberSecEval 4?
MAI-Thinking-1 is currently ranked first with 63.0%.
What does CyberSecEval 4 measure?
CyberSecEval 4 is an evaluation suite covering cybersecurity-related capabilities and risks of large language models. The insecure-code-generation tracks measure whether a model produces vulnerable code: the Instruct track presents coding requests designed to elicit known insecure patterns, while the Autocomplete track prompts the model with code context leading up to a known insecure pattern, with vulnerabilities detected via static analysis.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 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.