safety benchmark
SEC-bench Pro
SEC-bench Pro is a self-evolving software-security benchmark that measures agent bug hunting on critical, high-complexity systems. It instantiates validated vulnerabilities across the V8 and SpiderMonkey JavaScript engines as reproducible vulnerability-discovery and proof-of-concept-generation tasks with oracle-based validation.
Updated Aug 7, 2026
Published models3
Registry coverage3
MetricScore
EvidenceB
SEC-bench Pro leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | OP | 71.2% | 100.0% | 3 | C | |
| 2 | OP | 57.7% | 50.0% | 3 | C | |
| 3 | OP | 48.9% | 0.0% | 3 | C |
Score distribution
Top published rows on the benchmark's original scale.
SEC-bench Pro highlights
The top published results on this benchmark's own scale.
What is SEC-bench Pro?
Definition and scoring fields from the benchmark registry.
SEC-bench Pro is a self-evolving software-security benchmark that measures agent bug hunting on critical, high-complexity systems. It instantiates validated vulnerabilities across the V8 and SpiderMonkey JavaScript engines as reproducible vulnerability-discovery and proof-of-concept-generation tasks with oracle-based validation.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- SEC-bench Pro
- Modality
- text
- Primary category
- safety
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- sec-bench-pro|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about SEC-bench Pro.
Which model scores highest on SEC-bench Pro?
GPT-5.6 Sol is currently ranked first with 71.2%.
What does SEC-bench Pro measure?
SEC-bench Pro is a self-evolving software-security benchmark that measures agent bug hunting on critical, high-complexity systems. It instantiates validated vulnerabilities across the V8 and SpiderMonkey JavaScript engines as reproducible vulnerability-discovery and proof-of-concept-generation tasks with oracle-based validation.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
3 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.