AI Reliability Console

Monitor behavior, manage risk, and ship confidently

Evaluation Studio / Run

Support Assistant v3 - Pre-release

Support Assistant · gpt-4.1 · run_224

Back to RunsDecision WorkspacePublish Report
Decision
Pending
Confidence
68%
Overall Score
84
Release Gate
2.7%

Critical Failure Rate

Release GateRegressionCost EfficiencyDisagreement 13.4%

Policy citation failures are concentrated in refund edge-cases, with elevated cross-judge disagreement.

Derived metrics

Helpfulness: 82Information Density: 79Format Adherence: 94Coherence: 78

Decision metrics by intent

Release Gate

Decision: Pending

Critical Failure Rate: 2.7%

Pass Rate: 89.2%

Regression

Regression %: 2.1%

Result Variance %: 4.6%

Pass Rate: 89.2%

Cost Efficiency

Average Cost Per Run: $0.018

Total Cost of Bench: $37.8

Total Tokens (Input / Output): 1.82M / 0.94M

Policy pack: Regulated Support v4Evaluator versions lockedResidency: US-EastAudit evidence chain: active
Run Metric Timeline
Quality and risk evolution during execution.
Judge Score Breakdown
Per-metric score spread across evaluators.