▸ Uptime · 90d
99.984%▲
▸ API · p50
42ms
▸ Connect time
1.8s
▸ Incidents · 30d
2
Components
Operational
Degraded
Outage
Carriers
Last 24 hours
Twilio
Primary · US-East · STIR/SHAKEN A+
99.99%24h
Operational
Bandwidth
US-Central · STIR/SHAKEN A+
100%24h
Operational
Telnyx
Global · A
99.94%24h
Operational
Plivo
EU-West · A
99.87%24h
Elevated latency
Vonage
APAC · A
99.96%24h
Operational
Incident history
Plivo EU-West · elevated post-dial latency
May 12, 2026
14:08 UTC
14:08 UTC
Plivo reports an internal routing change is causing ~140ms additional post-dial delay for EU-West calls. Active calls unaffected. We've temporarily failed over EU-West outbound to Telnyx for new calls.
14:42 UTC · Monitoring Failover to Telnyx complete. EU-West connect time back to 1.6s avg. Watching for full Plivo resolution.
14:18 UTC · Investigating Confirmed with Plivo support. Issue is on their side; ETA 60–90 min. We're routing new EU-West calls through Telnyx.
14:08 UTC · Investigating Alerting triggered on EU-West post-dial p95.
AI transcription · degraded 14 minutes
May 4, 2026
09:24 UTC
09:24 UTC
A model deployment rolled out an updated tokenizer that increased live-coach latency by ~600ms for 14 minutes. Auto-rollback fired and traffic stabilized. Post-call summaries were unaffected.
09:38 UTC · Resolved Auto-rollback completed; latency restored to baseline.
09:31 UTC · Identified Tokenizer regression in v3.42. Initiating rollback.
09:24 UTC · Investigating Coach latency alerting tripped on p95.
Dashboard · slow loads in US-West
Apr 28, 2026
22:14 UTC
22:14 UTC
CloudFront cache miss rate spiked on the dashboard JS bundle in US-West for ~22 minutes after a deploy. TTFB rose from 80ms to 1.8s for new sessions. Active sessions unaffected.
22:36 UTC · Resolved Cache warm-up complete across edge locations.
22:22 UTC · Identified Stale invalidation on new bundle. Manually warming edges.
22:14 UTC · Investigating Customer reports of slow load. Reproduced internally.
Get incident updates
Email, Slack, or RSS — pick how we tell you something's wrong.