Model provider failures ongoing across Cartesia, Deepgram and Elevenlabs
Vapi reports degraded model provider service affecting Cartesia, Deepgram and ElevenLabs. Vapi reports that it sees improvement for Cartesia voices. Deepgram and ElevenLabs issues are still ongoing after 44 minutes. As a workaround, Vapi suggests Vapi V2 voices, xAI or MiniMax for voice, and AssemblyAI or Soniox for transcription.
State
Degraded
Started
Ongoing for
47 min
Worst impact
Degraded
Timeline
5 updatesDegraded
47 min 100%
Updates
Times in UTCAs posted by Vapi
We are seeing improvement for Cartesia voices. Deepgram and ElevenLabs issues are still ongoing.
See status pages for Deepgram and Elevenlabs for updates. https://status.elevenlabs.io/united-states/incidents/01M7QMQH05JXV3SKSM1H6PJZD2 https://status.deepgram.com/ Potential get out of trouble alternatives: Voice: Use Vapi V2 voices (such as Elliot or Clara), xAI, or MiniMax. ElevenLabs pronunciation dictionaries do not carry over and should not be added back after switching else, voice will default back to ElevenLabs. Transcriber: use AssemblyAI or Soniox for English, or Soniox for multilingual. If you have HIPAA or other data handling requirements, we recommend Cartesia or xAI voices.
See status pages for Cartesia, Deepgram and Elevenlabs for updates. https://status.elevenlabs.io/united-states/incidents/01M7QMQH05JXV3SKSM1H6PJZD2 https://status.deepgram.com/ Potential get out of trouble alternatives: Voice: Use Vapi V2 voices (such as Elliot or Clara), xAI, or MiniMax. ElevenLabs pronunciation dictionaries do not carry over and should not be added back after switching else, voice will default back to ElevenLabs. Transcriber: use AssemblyAI or Soniox for English, or Soniox for multilingual. If you have HIPAA or other data handling requirements, we recommend Cartesia or xAI voices.
See status pages for Cartesia, Deepgram and Elevenlabs for updates. https://status.elevenlabs.io/united-states/incidents/01M7QMQH05JXV3SKSM1H6PJZD2 https://status.deepgram.com/ For get out of trouble alternatives: - for Voices, consider xAI (Vapi V2) or Minimax - for English - consider AssemblyAI or Sonion; for multi-lingual Soniox
See status pages for Cartesia, Deepgram and Elevenlabs for updates. For get out of trouble alternatives: - for Voices, consider xAI (Vapi V2) or Minimax - for English - consider AssemblyAI or Sonion; for multi-lingual Soniox
Affected components
3Vapi in the last 90 days
Full historyIncidents in 90 days
10
Days with an incident
13
Median duration
25 min
This incident
1.9× the median
Other Vapi incidents
All incidentsVapi·Degraded Service: Insights API and Boards Inaccessible
Vapi reports that the Insights API and Dashboard boards did not work. Calls and structured outputs were not affected. Vapi resolved the issue on 11 November 2026, about 43 hours after it started.
Vapi·Dashboard is unavailable
The issue is resolved and all services have returned to normal.
Vapi·docs.vapi.ai is unavailable
The issue has been resolved upstream and documentation is available once again, we'll continue monitoring to ensure stability.
Vapi·Degraded performance: SIP BYO trunk creation
BYO SIP registration has been re-enabled during scheduled maintenanc and we were able to successfully place outbound calls
Vapi·SIP Server downtime
There will be reactive maintenance on SIP outbound dialing infrastructure, US region Status: Scheduled What is affected: Our SIP infrastructure provider will upgrade the web server/API that controls outbound dialing, as part of remediating the root cause of today's incident. Inbound calls are not affected. When: Tuesday, October 6, 21:00–21:15 PT (04:00–04:15 UTC, Oct 7) What to expect: This work is not expected to be customer-impacting or noticeable. Outbound calls may briefly stall during the window; any impact should last less than 10 minutes. Action needed: None
Follow Vapi
Your coding agent knows when Vapi is down, and tells you when a failure is theirs, not yours. You can also get Vapi incidents in a feed reader or in Slack.
Install the outages.live MCP server, so that when something breaks you can tell if the cause is a provider we depend on and not my code. It reads the official status pages of cloud, AI, payment and developer providers and gives you their incidents as MCP tools.
Add it as a remote MCP server named "outages" at https://outages.live/mcp (Streamable HTTP, no authentication), with your CLI command or your MCP config file.
Then:
1. Make sure it connects and shows the tools get_provider, list_feed and send_feedback.
2. Call get_provider with "vapi" and tell me if Vapi has a problem right now.
3. From now on, when a call to a third-party service fails or becomes slow, call get_provider for that service before you change my code.Paste this into any coding agent. It adds the server with its own MCP setup.
https://outages.live/feed.xml?providers=vapi/feed subscribe https://outages.live/feed.xml?providers=vapiFAQ
Go from tracking outages to fixing production issues with Polylane.
Try Polylane