Agentic Person Search (Temporal Reasoning) on Track 3 Temporal
100TWSOracle
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| OracleBackbone=Oracle2026.04 | 100 | 100 | 100 | 1.88 | |
| ARGOS AgentBackbone=GPT-5.22026.04 | 59 | 88.2 | 65.8 | 3.91 | |
| ARGOS AgentBackbone=GPT-4o2026.04 | 56.7 | 80.6 | 60.8 | 3.91 | |
| ARGOS AgentBackbone=GPT-5-mini2026.04 | 55.6 | 88 | 60.5 | 4.4 | |
| ARGOS AgentBackbone=Cl. Sonnet 42026.04 | 54.8 | 83.6 | 59.5 | 4.25 |