NUWA Evidence record
GPT-5.6 Sol + Codex 0.144.1
AgentCyberRange · Evidence record
Original source charts and materials
Reported results
Source order · reported configurations
Select systems
Web · Level-0 Pass@3
0–100%
GPT-5.6 Sol + Codex 0.144.1Post · Level-0 Pass@3
0–100%GPT-5.6 Sol + Codex 0.144.1Values describe the measured outcome under the source protocol, not a general safety score.
Level-0, Pass@3. Harness versions differ; model revisions and budgets are incompletely reported. Evaluation date not specified; publication / record dates are shown below.
Evaluation conditions and system configurations
- GPT-5.6 Sol + Codex 0.144.1
- The site describes 110 web vulnerabilities and 156 hosts in 8 ranges. Per-metric independent trial counts are not recorded. · Website's Level-0 web and post-exploitation tracks
- modelVersion: not specified · harness: Codex 0.144.1 · tools: Project-provided web and enterprise-like range environments · budget: not specified
| System / configuration | Web · Level-0 Pass@3 | Post · Level-0 Pass@3 | Publication / verification |
|---|---|---|---|
| GPT-5.6 Sol + Codex 0.144.1Codex 0.144.1 | 35.37% | 74.10% | Source checked2026-10-24 |
Method and scope
Task success across web exploitation and post-exploitation settings; configurations must be checked per submission.
- Evidence type
- Benchmark
- Assessment subject
- AI agent completing cyber-range tasks
- Primary dimension
- Capability
- Evidence snapshot
- 2026-10-29.1
Primary dimension follows the source’s measured construct. Metric-level interpretations may differ.
Source finding
The project reports these values for the named model and agent version.
NUWA interpretation
These are submission results from two distinct task tracks. Pass@3 is retained as reported; no cross-track or cross-study score is computed.
Limitations
Changing task sets and submitted agent configurations prevent unrestricted comparison.
The captured website does not give evaluation dates or complete budgets for these rows. Model aliases may change. This is a dated source snapshot, not an independent NUWA reproduction.
Related research questions
Cybersecurity / Cyber task execution →
The source directly measures the stated construct under its protocol.
References
- [1] AgentCyberRange: Evaluate Agent Cyberattack Capabilities in Enterprise-like Cyber Ranges ; Official leaderboard: Level-0 Pass@3; arrays QR (web), $R (post), version map ez
- [2] AgentCyberRange: Evaluating autonomous cyber capabilities over long-horizon tasks ; Research introduction: Fudan–Whitzard team