Tavus Outage History

49 incidents recorded over the last 365 days, 48 of them resolved. Durations are measured from when PulsAPI first saw the problem to when it cleared, which is usually longer than the vendor's own figure.

Live Tavus status99.50% uptime history

Every Recorded Tavus Incident

Tavus incidents, newest first, with date, duration and status.
DateIncidentDurationStatus
Aug 13, 2026Elevated start up times for some conversationsWhen one of our GPU providers performed maintenance, we asked to migrate traffic to a different cluster beforehand to avoid disruptions. During the migration back, the provider encountered an issue that led to some…-19 minResolved
Aug 5, 2026Intermittent Response Issues with Cartesia TTSCartesia has fully restored service, and conversations using Cartesia TTS are responding normally.14 minResolved
Aug 5, 2026Some Custom Replica Creations failing due to upstream Infra Provider IssueThis incident has been resolved.1h 15mResolved
Aug 3, 2026Some Conversations Failing to StartAll conversations are starting normally.26 minResolved
Aug 1, 2026Cartesia IncidentThis incident has been resolved.18 minResolved
Jul 30, 2026Increased conversation latency using the Gemma 4 modelConversation latency for the Gemma 4 model has returned to normal, and all traffic has been routed.41 minResolved
Jul 29, 2026Some conversations are fail to startCartesia has restored service, and conversations configured to use Cartesia for TTS are starting successfully again.14 minResolved
Jul 24, 2026Some conversations are unable to startConversation creation temporarily failed for approximately 10 minutes. The issue has been resolved, and conversation creation is operating normally.13h 10mResolved
Jul 15, 2026OSS LLM (tavus-oss) elevated latencyUncovered and rolled back related change - OSS LLM (tavus-oss) traffic is healthy again.2h 29mResolved
Jul 6, 2026Delaying Joining on Google Meet InvitesThis incident has been resolved.57 minResolved
Jun 17, 2026Replica Join Failures for New ReplicasA fix has been deployed and this incident is now resolved. A very limited subset of conversations using newly trained replicas were affected by join failures. We identified the root cause and do not expect any further…1h 49mResolved
Jun 16, 2026Async video generation and replica training are processing slower than expectedThis issue has been resolved, and all existing replica trainings and video generations are now processing as expected. Our team is continuing to monitor all runs.13 minResolved
May 8, 2026Some conversations not startingThis incident has been resolved.26 minResolved
Apr 30, 2026Degraded Performance Across Multiple ServicesThis incident has been resolved.5h 13mResolved
Apr 29, 2026Increased error rates with perception modelsThe incident has been resolved, and Raven is now working as expected.5h 8mResolved
Apr 17, 2026Intermittent Call Connection IssuesCloudflare has implemented a fix and downstream services are returning to normal service3h 45mResolved
Apr 14, 2026Replica takes longer to join in some conversationsThis has been resolved.36 minResolved
Apr 10, 2026Elevated Error Rate When Joining ConversationsThis incident has been resolved14 minResolved
Apr 5, 2026Conversations not properly initiatingThis incident has been resolved.7 minResolved
Apr 4, 2026Some conversations slow to bootThis incident has been resolved.19 minResolved
Mar 26, 2026Some conversations not initiatingWe experienced a brief disruption due to a configuration change that impacted conversation routing, causing some conversations to stall or take longer to start. The issue has been resolved and all systems are now…32 minResolved
Mar 25, 2026Some conversations are encountering increased startup timesThis incident has been resolved.14 minResolved
Mar 10, 2026Some conversations took longer time startWe detected several conversations with start times exceeding 10 seconds. We have diverted traffic and are continuing to monitor the situation.0 minResolved
Mar 6, 2026Some conversations not starting upWe have identified an issue with one of our GPU providers and shifted traffic away from them-18 minResolved
Feb 17, 2026List Replicas EndpointAPI is back to operational17 minResolved
Feb 12, 2026Conversations slow to joinThis incident has been resolved.24 minResolved
Jan 23, 2026Conversations are seeing delayed initiationThis incident has been resolved.1h 41mResolved
Jan 21, 2026Some conversations not initiating due to webRTC issueThis incident has been resolved.3h 11mResolved
Jan 15, 2026Some conversations not initiatingThis incident has been resolved.1h 1mResolved
Jan 7, 2026Increased wait times for replica to join conversationsThis incident has been resolved.2h 39mResolved
Dec 18, 2025Temporary Video Generation OutageThis incident has been resolved.3 minResolved
Dec 6, 2025Elevated error rates with Llama-4 due to downstream provider errorsThis incident has been resolved.53 minResolved
Dec 2, 2025Llama-4 Elevated Error RatesThis incident has been resolved.1h 22mResolved
Dec 2, 2025Replica Training and Video Generation Audio IssueThis incident has been resolved2h 37mResolved
Nov 20, 2025Intermittent errors on Conversation CreationOur firewall identified anomalous traffic patterns and the systems are now stabilized.16 minResolved
Nov 20, 2025LiveKit Service DisruptionThis incident has been resolved.4h 17mResolved
Nov 18, 2025Conversation creation failingServices are back to normal operations. Let us know if you still see issues support@tavus.io .6h 18mResolved
Nov 13, 2025Upstream LLM Provider DisruptionThis incident has been resolved.4h 0mResolved
Nov 12, 2025Experiencing intermittent log in issuesThis incident has been resolved.20 minResolved
Nov 12, 2025Increased utterance to utterance latency for tavus-llamaThis incident has been resolved.1h 30mResolved
Nov 11, 2025Conversation Start Failures (GPU Provider Degradation)This incident has been resolved.3h 7mResolved
Nov 5, 2025Conversation creation failingThis outage was caused by a GPU provider outage, and traffic has now been redirected. All conversations are properly initiating and services are fully restored. Our team will continue to closely monitor.10 minResolved
Oct 26, 2025Scheduled Maintenance - Sun, Oct 26, 2025 at 2:00–3:00 AM PDT (09:00–10:00 UTC)The scheduled maintenance has been completed.30 minScheduled
Oct 20, 2025Some Conversations Not Joined By Replica due to Ongoing AWS outageThis incident has been resolved.13h 5mResolved
Oct 20, 2025AWS outage causing Conversation Creation TimeoutsSome conversation creation requests were impacted by the AWS us-east-1 incident and resulted in a 504 Gateway Timeout. This affected CVI. The incident lasted from 10:00 UTC and lasted until 10:20 UTC.2h 30mResolved
Oct 14, 2025Cartesia outage resulting in text to speech failuresThis incident has been resolved.19h 46mResolved
Oct 5, 2025Delayed Video Generation CallbacksSome video generation callbacks didn't properly deliver between October 3–5 as the callback jobs remained in an "in-progress" state and did not send as expected. The issue has been fully fixed now, and all affected…0 minResolved
Sep 30, 2025Video List RegressionThis incident has been resolved34 minResolved
Sep 20, 2025Some CVI Conversations failing to startThis incident has been resolved.11 minResolved

How to Read This Tavus Incident Log

Each row is an incident PulsAPI observed, not a summary written afterwards. The duration is wall-clock time between the first failing check and the first clean one, so it includes the window before Tavus acknowledged anything. Vendor post-mortems typically measure from acknowledgement, which is why their numbers are usually shorter.

Incidents still open have no duration yet and are listed as ongoing rather than being given a running total. The archive covers the last 365 days; anything older has aged out of the window rather than never having happened.

The longest single Tavus outage in this window ran 19h 46m, against a mean recovery of 2h 19m. If you depend on Tavus in a customer-facing path, the longest figure is the one to design around. The mean is what happens on a normal bad day; the maximum is what happens on the worst one.