Observability overview
Dashboard
Latency, cost, usage, errors, and cancellation health from real inference logs.
Overall health
Total requests
0
Gateway calls observed
Success rate
0%
0 of 0 completed
P95 latency
-
Tail request duration
Estimated cost
$0
Provider cost estimate
Inference performance
P50 latency
-
Median request duration
P50 TTFT
-
Median time to first token
P95 TTFT
-
Tail time to first token
Total tokens
0
Input plus output tokens
Resilience
Retries
0
Retry attempts
Fallbacks
0
Alternate provider routes
Dead letters
0
Worker messages requiring review
Error rate
0%
Provider or request failures
Product activity
Replays
0
Linked replay traces
Comparisons
0
Comparison runs
Cancellations
0%
User-stopped streams
PII redactions
0
Persisted redaction events
Latency over time
Average latency by day
No inference data yet. Send a chat message to populate this chart.
Provider/model usage split
Requests by configured target
No inference data yet. Send a chat message to populate this chart.
Status distribution
Completed, failed, and cancelled requests
No inference data yet. Send a chat message to populate this chart.
Token usage
Daily token volume from logs
No inference data yet. Send a chat message to populate this chart.