safety benchmark
Capture-the-Flag Challenges (Internal)
Capture-the-Flag Challenges is OpenAI's internal expansion of competitive, professional-level cybersecurity capture-the-flag tasks used to evaluate vulnerability identification and exploitation capability under the Preparedness Framework.
Updated Aug 7, 2026
Published models3
Registry coverage3
MetricScore
EvidenceB
Capture-the-Flag Challenges (Internal) leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | OP | 96.7% | 100.0% | 3 | C | |
| 2 | OP | 91.8% | 50.0% | 3 | C | |
| 3 | OP | 85.2% | 0.0% | 3 | C |
Score distribution
Top published rows on the benchmark's original scale.
Capture-the-Flag Challenges (Internal) highlights
The top published results on this benchmark's own scale.
What is Capture-the-Flag Challenges (Internal)?
Definition and scoring fields from the benchmark registry.
Capture-the-Flag Challenges is OpenAI's internal expansion of competitive, professional-level cybersecurity capture-the-flag tasks used to evaluate vulnerability identification and exploitation capability under the Preparedness Framework.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- Capture-the-Flag Challenges (Internal)
- Modality
- text
- Primary category
- safety
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- capture-the-flag-challenges|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about Capture-the-Flag Challenges (Internal).
Which model scores highest on Capture-the-Flag Challenges (Internal)?
GPT-5.6 Sol is currently ranked first with 96.7%.
What does Capture-the-Flag Challenges (Internal) measure?
Capture-the-Flag Challenges is OpenAI's internal expansion of competitive, professional-level cybersecurity capture-the-flag tasks used to evaluate vulnerability identification and exploitation capability under the Preparedness Framework.
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.