Alerts

Threshold monitors on evaluation and trace metrics.

Rule	Metric	Condition	Scope
Pass rate drops below 70%	Pass rate	< 70%	global
Latency exceeds 3s	Latency	> 3000ms	global
Trace cost exceeds $0.01	Cost	> $0.01	global
Error rate exceeds 10%	Error rate	> 10%	global

Trace cost exceeds $0.01Jul 15, 09:30 PM

Trace "summarize.ticket" (gpt-4o-mini) cost $0.0115 — above the $0.01 threshold.

Latency exceeds 3sJul 13, 09:30 PM

Trace "summarize.ticket" (claude-sonnet-5) took 3544ms to complete.

Latency exceeds 3sJul 10, 09:30 PM

Trace "chat.completion" (gpt-4o-mini) took 3891ms to complete.

Latency exceeds 3sJul 9, 09:30 PM

Trace "rag.retrieve_and_generate" (gemini-2.5-pro) took 3545ms to complete.

Latency exceeds 3sJul 8, 09:30 PM

Trace "summarize.ticket" (claude-haiku-4-5) took 3781ms to complete.

Pass rate drops below 70%Jul 7, 09:30 PM

Run "Support QA — claude-sonnet-5" passed only 13% of test cases.

Latency exceeds 3sJul 6, 09:30 PM

Trace "summarize.ticket" (claude-sonnet-5) took 4137ms to complete.

Trace cost exceeds $0.01Jul 6, 09:30 PM

Trace "summarize.ticket" (claude-sonnet-5) cost $0.0114 — above the $0.01 threshold.

Trace cost exceeds $0.01Jul 6, 09:30 PM

Trace "chat.completion" (gpt-4o-mini) cost $0.0118 — above the $0.01 threshold.

Trace cost exceeds $0.01Jul 2, 09:30 PM

Trace "chat.completion" (gemini-2.5-pro) cost $0.0118 — above the $0.01 threshold.