Vulnerability Detection on 28 CVEs (Total)
15True Positives (TPs)ANTAEUS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ANTAEUSLLM=Claude Opus 4.72026.07 | 15 | 1,345 | |
| ANTAEUSLLM=GPT G5.42026.07 | 12 | 2,918 | |
| Function-level baselineLLM=Claude2026.07 | 5 | 1,197 | |
| Agentic baselineLLM=Claude Opus 4.8, Prioritization Ranking=false2026.07 | 5 | 190 | |
| Agentic baselineLLM=Claude Opus 4.8, Prioritization Ranking=true2026.07 | 4 | 367 | |
| Function-level baselineLLM=GPT2026.07 | 4 | 4,971 | |
| Agentic baselineLLM=Claude Opus 4.7, Prioritization Ranking=false2026.07 | 3 | 545 | |
| Agentic baselineLLM=Claude Opus 4.7, Prioritization Ranking=true2026.07 | 3 | 967 |