Evaluation quality matrix
Model scores across production quality dimensions.
CorrectnessFaithfulnessRelevanceSafetyStyle
Agent observability
Trace behavior, model quality, latency and cost across production agents.
Span-level latency and tool execution paths.
Model scores across production quality dimensions.
Automated production policy coverage.
88%
quality pass
Traffic allocation by task and provider.
Pending regressions and production review sets.
5 production traces in the current model view.
| Trace | Model | Latency | Tokens | Quality | State |
|---|---|---|---|---|---|
support-resolution-agent tr_0849 · Support AI · 12s ago | gpt-5 | 842ms | 12,840 | 94% | Healthy |
invoice-audit-workflow tr_0848 · Finance AI · 34s ago | claude | 1280ms | 18,420 | 87% | Slow |
catalog-enrichment-batch tr_0847 · Commerce AI · 1m ago | gemini | 618ms | 9,600 | 91% | Healthy |
lead-qualification-agent tr_0846 · Revenue AI · 3m ago | gpt-5 | 1540ms | 22,180 | 71% | Failed |
contract-risk-review tr_0845 · Legal AI · 6m ago | claude | 974ms | 14,620 | 96% | Healthy |