Q4 Fraud Impact Report - Email & Ticket InvestigationRisk & Fraud
How each model performed on this task. Every chart states its N, and the reward carries a 95% confidence interval once there is more than one trial to measure.
Core metrics
Check Pass %
Mean share of the task's checks passing, averaged over trials. N = 10 trials per model.
54%
Claude Opus 4.8
48%
GPT-5.5
0%
Gemini 3.1 Pro
0%
Gemini 3.5 Flash
Extended metrics
Median Time
Wall-clock minutes, median across trials. N = 10 trials per model.
3m
GPT-5.5
3m
Claude Opus 4.8
1m
Gemini 3.1 Pro
1m
Gemini 3.5 Flash
Tool Calls
Tool invocations, median across trials — counted from the trajectories. N = 10 trials per model.
72
GPT-5.5
35
Claude Opus 4.8
13
Gemini 3.1 Pro
1
Gemini 3.5 Flash
Verifier × model pass rate
How often each verifier passed, per model. Green is passing, red is failing — a red band across a row is a verifier nothing clears.
| Verifier | GPT-5.5n=10 | Claude Opus 4.8n=10 | Gemini 3.1 Pron=10 | Gemini 3.5 Flashn=10 |
|---|---|---|---|---|
| checks (11) | ||||
| found_fraudulent_activity_thread | 40% | 40% | 0% | 0% |
| found_unauthorized_charges_thread | 80% | 50% | 0% | 0% |
| q4_fraud_impact_channel_in_database | 100% | 100% | 0% | 0% |
| slack_lists_ticket_11533 | 0% | 40% | 0% | 0% |
| slack_lists_ticket_20644 | 20% | 40% | 0% | 0% |
| avoided_deletions | 100% | 100% | 100% | 100% |
| avoided_ticket_status_changes | 100% | 100% | 100% | 100% |
| checked_conversations_ticket_11533 | 20% | 10% | 0% | 0% |
| checked_conversations_ticket_20644 | 20% | 0% | 0% | 0% |
| found_fraud_ticket_11533 | 20% | 10% | 0% | 0% |
| found_fraud_ticket_20644 | 20% | 0% | 0% | 0% |
Failure modes on this task
40 tagged verifier failures, by the stage they were assigned to. Higher is worse.
| Failure stage | GPT-5.5 | Claude Opus 4.8 | Gemini 3.1 Pro | Gemini 3.5 Flash |
|---|---|---|---|---|
| Answer1 mode | 10 | 10 | 10 | 10 |