Audit manual-review check deposit release latency & stranded fundsDeposits & Funding
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.
98%
GPT-5.5
74%
Claude Opus 4.8
66%
Gemini 3.1 Pro
1%
Gemini 3.5 Flash
Extended metrics
Median Time
Wall-clock minutes, median across trials. N = 10 trials per model.
7m
Gemini 3.5 Flash
4m
Gemini 3.1 Pro
3m
Claude Opus 4.8
3m
GPT-5.5
Tool Calls
Tool invocations, median across trials — counted from the trajectories. N = 10 trials per model.
44
Gemini 3.1 Pro
36
Claude Opus 4.8
26
GPT-5.5
18
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 (13) | ||||
| gold_accounts_multi_stranded | 100% | 80% | 70% | 0% |
| gold_affected_accounts | 100% | 80% | 70% | 0% |
| gold_max_account_stranded_usd | 100% | 80% | 70% | 0% |
| gold_oldest_stuck_age_days | 100% | 80% | 70% | 0% |
| gold_risk_hold_count | 100% | 80% | 90% | 0% |
| gold_silent_ocr_count | 100% | 90% | 70% | 0% |
| gold_stuck_deposit_count | 100% | 90% | 70% | 0% |
| gold_stuck_dollars_held | 100% | 80% | 70% | 0% |
| rubric_root_cause | 80% | 0% | 0% | 0% |
| rubric_uses_all_connectors | 100% | 100% | 100% | 10% |
| scopes_stranded_set_and_cohorts | 100% | 50% | 50% | 0% |
| gold_risk_hold_dollars | 100% | 80% | 90% | 0% |
| process_floor | 100% | 100% | 100% | 100% |