Trace-level detection on TRACE
76.4APMonitor
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MonitorTPC=10, Backbone Model=Qwen3.52026.04 | 76.4 | — | |
| MeerkatTPC=10, Backbone Model=Qwen3.52026.04 | 76 | -0.004 | |
| MeerkatTPC=25, Backbone Model=GLM-52026.04 | 73 | 0.018 | |
| MeerkatTPC=10, Backbone Model=GLM-52026.04 | 72.1 | 0.036 | |
| MonitorTPC=10, Backbone Model=GPT-5.4m2026.04 | 71.9 | — | |
| MonitorTPC=25, Backbone Model=GLM-52026.04 | 71.1 | — | |
| MeerkatTPC=50, Backbone Model=Qwen3.52026.04 | 69 | 0.018 | |
| MonitorTPC=10, Backbone Model=GLM-52026.04 | 68.5 | — | |
| MeerkatTPC=10, Backbone Model=GPT-5.4m2026.04 | 67.3 | -0.047 | |
| MonitorTPC=25, Backbone Model=Qwen3.52026.04 | 67.3 | — | |
| MonitorTPC=50, Backbone Model=Qwen3.52026.04 | 67.2 | — | |
| Naive AgentTPC=10, Backbone Model=Qwen3.52026.04 | 64 | — | |
| MeerkatTPC=25, Backbone Model=Qwen3.52026.04 | 62.6 | -0.047 | |
| MeerkatTPC=25, Backbone Model=GPT-5.4m2026.04 | 60.4 | 0.013 | |
| MonitorTPC=25, Backbone Model=GPT-5.4m2026.04 | 59 | — | |
| MeerkatTPC=50, Backbone Model=GLM-52026.04 | 59 | 0.007 | |
| MonitorTPC=50, Backbone Model=GLM-52026.04 | 58.3 | — | |
| MonitorTPC=50, Backbone Model=GPT-5.4m2026.04 | 52.4 | — | |
| MonitorTPC=100, Backbone Model=Qwen3.52026.04 | 52.1 | — | |
| MeerkatTPC=50, Backbone Model=GPT-5.4m2026.04 | 51.5 | -0.009 | |
| MeerkatTPC=100, Backbone Model=GLM-52026.04 | 49.1 | 0 | |
| MonitorTPC=100, Backbone Model=GLM-52026.04 | 49.1 | — | |
| MeerkatTPC=100, Backbone Model=Qwen3.52026.04 | 48.1 | -0.04 | |
| Naive AgentTPC=25, Backbone Model=Qwen3.52026.04 | 39 | — | |
| Naive AgentTPC=50, Backbone Model=Qwen3.52026.04 | 35.9 | — | |
| MonitorTPC=100, Backbone Model=GPT-5.4m2026.04 | 33 | — | |
| MeerkatTPC=100, Backbone Model=GPT-5.4m2026.04 | 31.4 | -0.016 | |
| Naive AgentTPC=100, Backbone Model=Qwen3.52026.04 | 19.6 | — |