Monitoring
Latency, cost and health for every agent in Halvard Labs.
Requests / min
35.9
▲ 11%Error rate
1.1%
▼ 0.2%p95 latency
2.38s
▼ 4%Tokens / min
142k
▲ 8%Latency
p50 and p95 across all agents · last 24 hours
- p50
- p95
Token usage by model
Last 14 days
945Mtokens
- Atlas-4 Turbo44%
- Atlas-428%
- Atlas-4 Mini20%
- Lyra-2 Embed8%
Cost per day
Model and tool spend, last 14 days
- Total
- $968.20
- Daily avg
- $69.16
Agent health
Last 24 hours · runs per hour trend
| Agent | Status | Runs | Error rate | p95 | Cost | Trend |
|---|---|---|---|---|---|---|
| Support TriageAtlas-4 Turbo | Live | 18,420 | 1.4% | 2.1s | $41.30 | |
| Invoice ReconcilerAtlas-4 | Live | 3,912 | 3.9% | 5.9s | $22.70 | |
| Sales ResearcherAtlas-4 Turbo | Live | 1,268 | 5.6% | 11.2s | $18.90 | |
| Code ReviewerAtlas-4 | Paused | 0 | — | — | $0.00 | No runs |
| Onboarding GuideAtlas-4 Mini | Live | 9,604 | 0.8% | 980ms | $6.40 | |
| Contract AnalystAtlas-4 | Failing | 214 | 28.7% | 14.3s | $4.10 |
System status
Daily uptime for the last 90 days
Partial degradation
- APIPublic REST and streaming endpoints99.98% uptimeOperationalJul 17Today
- WorkersRun execution and tool sandbox99.91% uptimeDegradedJul 17Today
- Vector storeKnowledge indexes and retrieval100% uptimeOperationalJul 17Today
- WebhooksOutbound event delivery99.95% uptimeOperationalJul 17Today
OperationalDegradedOutage