Net Institutional Fee RevenueRevenue & Fees
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.
93%
GPT-5.5
92%
Gemini 3.1 Pro
84%
Claude Opus 4.8
49%
Gemini 3.5 Flash
Extended metrics
Median Time
Wall-clock minutes, median across trials. N = 10 trials per model.
7m
Gemini 3.5 Flash
7m
Gemini 3.1 Pro
5m
Claude Opus 4.8
3m
GPT-5.5
Tool Calls
Tool invocations, median across trials — counted from the trajectories. N = 10 trials per model.
59
GPT-5.5
53
Gemini 3.5 Flash
52
Gemini 3.1 Pro
49
Claude Opus 4.8
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 (44) | ||||
| answer_card_replacement_family_net | 100% | 100% | 100% | 80% |
| answer_check_issuance_family_net | 100% | 100% | 100% | 80% |
| answer_fee_credits_and_refunds | 70% | 60% | 100% | 80% |
| answer_fee_subject_busiest_requester | 80% | 50% | 100% | 0% |
| answer_fee_subject_busiest_requester_ticket_count | 80% | 50% | 100% | 0% |
| answer_fee_subject_distinct_requester_count | 80% | 50% | 100% | 0% |
| answer_fee_subject_open_count | 100% | 100% | 100% | 0% |
| answer_fee_subject_open_or_pending_count | 100% | 100% | 100% | 0% |
| answer_fee_subject_pending_count | 100% | 100% | 100% | 0% |
| answer_fee_subject_ticket_count | 100% | 100% | 100% | 0% |
| answer_fee_subject_unresolved_created_2024_or_later_count | 100% | 100% | 100% | 0% |
| answer_fee_subject_unresolved_created_before_2024_count | 100% | 100% | 100% | 0% |
| answer_foreign_transaction_family_net | 100% | 100% | 100% | 80% |
| answer_gross_collected_fees | 100% | 100% | 100% | 70% |
| answer_highest_refund_event_rate_family | 100% | 100% | 100% | 80% |
| answer_highest_refund_event_rate_pct | 70% | 100% | 100% | 80% |
| answer_highest_refund_rate_family | 100% | 100% | 100% | 80% |
| answer_highest_refund_rate_pct | 100% | 100% | 100% | 80% |
| answer_jira_realized_as_nominal_issue | 100% | 100% | 100% | 90% |
| answer_jira_realized_as_posted_issue | 100% | 100% | 100% | 90% |
| answer_largest_peak_over_refund_by_family_account | 100% | 100% | 100% | 70% |
| answer_largest_peak_over_refund_by_family_amount | 100% | 100% | 100% | 70% |
| answer_largest_peak_over_refund_family | 100% | 100% | 100% | 70% |
| answer_net_fee_revenue | 70% | 60% | 100% | 70% |
| answer_oon_atm_family_net | 100% | 60% | 100% | 80% |
| answer_oon_atm_partial_collection_shortfall | 100% | 100% | 100% | 70% |
| answer_peak_month_net_fee_revenue | 100% | 60% | 100% | 70% |
| answer_peak_month_top_fee_family | 100% | 100% | 100% | 70% |
| answer_peak_net_revenue_month | 100% | 60% | 100% | 70% |
| answer_premium_family_net | 100% | 100% | 100% | 80% |
| answer_temporary_over_refund_account_family_pair_count | 100% | 60% | 100% | 70% |
| answer_total_peak_over_refund_exposure_by_family | 100% | 60% | 100% | 70% |
| answer_wire_family_net | 70% | 100% | 100% | 80% |
| jira_followup_ticket_exists | 100% | 100% | 0% | 10% |
| jira_followup_ticket_has_evidence | 60% | 0% | 0% | 0% |
| slack_close_channel_exists | 100% | 100% | 100% | 10% |
| slack_close_summary_has_required_figures | 70% | 10% | 0% | 0% |
| slack_close_summary_is_pinned | 100% | 100% | 100% | 10% |
| floor_confluence_fee_policy | 100% | 100% | 100% | 100% |
| floor_jira_followup_write | 100% | 100% | 0% | 10% |
| floor_jira_lineage_reads | 100% | 100% | 100% | 100% |
| floor_sql_investigation | 100% | 100% | 100% | 100% |
| trace_complete_scope_and_netting | 70% | 60% | 100% | 80% |
| trace_reconciled_posted_amount | 60% | 90% | 100% | 90% |