Sink Detection on 8,262 candidate sinks (54 exploitable, 8,208 non-exploitable) (full)
52True Positives (TP)GONDAR
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| GONDARBackbone=gpt-52026.04 | 52 | 331 | 2 | 7,877 | 353.39 | 13.6 | 96.3 | 23.8 | 96 | |
| GONDARBackbone=gpt-oss-120b2026.04 | 52 | 692 | 2 | 7,516 | 1.74 | 7 | 96.3 | 13 | 91.6 | |
| GONDARBackbone=sonnet-4.52026.04 | 51 | 304 | 3 | 7,904 | 751.93 | 14.4 | 94.4 | 24.9 | 96.3 | |
| GONDARBackbone=gpt-5-nano2026.04 | 51 | 413 | 3 | 7,795 | 4.6 | 11 | 94.4 | 19.7 | 95 | |
| GONDARBackbone=flash-lite2026.04 | 51 | 273 | 3 | 7,935 | 40.99 | 15.7 | 94.4 | 27 | 96.7 | |
| GONDARBackbone=glm-52026.04 | 51 | 309 | 3 | 7,899 | 19.32 | 14.2 | 94.4 | 24.6 | 96.2 | |
| GONDARBackbone=gemini-2.5-pro2026.04 | 50 | 308 | 4 | 7,900 | 123.74 | 14 | 92.6 | 24.3 | 96.2 | |
| SpotBugs2026.04 | 20 | 244 | 34 | 7,964 | 1.13 | 7.6 | 37 | 12.6 | 97 | |
| CodeQL2026.04 | 14 | 215 | 40 | 7,993 | 1.13 | 6.1 | 25.9 | 9.9 | 97.4 |