Hi Vapi, All my calls are ending with this reason. During the call, the assistant doesnt speak nor respond to anything that the customer verbalize. Here are some call ID's so you can debug: Org ID: 6d7e825c-c695-40e7-b299-383905a3745b Call IDS:019e7397-dd13-7000-9ccc-ffaa28b4eb99 ; 019e7397-51ce-766d-8afc-596e93c57b9f Org ID: d1f49a88-3e89-445c-b5df-4590fbab9ea6 Call IDS: 019e739a-db3f-7666-bfa1-7cf768744e1a ; 019e7385-b04a-766f-b98b-25a39c042047 Im only providing info about these 2 org IDS but this is happening in all orgs i'm admin. All orgs are connected to the same Azure API Key. In Azure platform, we dont see any problem or limitation with usage limits reached. Can you help us solving this please? If any additional details are needed, let me know! Thanks, its urgent!

Hey! Root cause confirmed: Azure's West US 2 region is down right now due to a power event in their datacenter. Your Azure OpenAI deployment lives in that region. That's why every call ends silently. Vapi is sending the request, Azure just isn't responding. Since this is urgent for you, here's the fastest way to get calls working again while Azure recovers: Switch your assistants to use OpenAI directly instead of Azure OpenAI. In the Vapi dashboard, go to Provider Keys, add an OpenAI API key, then update your assistant's model config to use OpenAI. Test one call and it should work immediately. Once Azure confirms full recovery for West US 2, you can switch everything back to Azure with no other changes. Azure's latest update was at 12:08 UTC and they're actively working on it, but there's no ETA yet. Let me know if you hit any issues with the workaround and I'll walk you through it. Regards, Chiranjeet Vapi Support

Hey @Chiranjeet Mishra ! Thank you very much for the quick update. We confirm that switching model resolved the problem. We will wait for azure update since we are now monitoring the url: https://azure.status.microsoft/pt-pt/status Thank you very much for fast answer!

@User This issue affected hundreds of our inbound healthcare calls last night during a ~3hr window, with many failing silently. Please can someone from the team get in touch with me to help ensure monitoring and fallbacks are properly configured in our orgs to reduce the chances of this happening moving and lessen the impact if it does moving forward.

Hey Louis, I'm really sorry about the impact this had on your healthcare calls. Silent failures on inbound calls in a healthcare context is serious, and I want to make sure we handle this properly. Root cause: this was caused by an Azure West US 2 regional outage triggered by a power event in their datacenter (started 04:27 UTC on May 29). Your Azure OpenAI deployment was in that region. Every LLM request timed out with no response, so calls connected but the assistant stayed silent. Azure has since restored the region, so calls on that setup should be working normally now. For monitoring going forward, the most reliable thing you can do today is watch your call.ended webhook events for an endedReason of

pipeline-error-azure-openai-llm-failed

. When that pattern spikes, it means your LLM provider is having trouble. Catching it there lets your team react quickly instead of waiting for customer complaints. For resilience, the practical approach is to keep a backup assistant configuration ready using a different LLM provider (for example, OpenAI directly, or a second Azure OpenAI deployment in a different region like East US). Switching in Vapi is fast when you have the config ready ahead of time. On your request for someone to walk through this with you properly: absolutely, this deserves more than a ticket. I'm flagging this internally so the right person from our team can reach out to set up time with you. Could you share your email address and the best way to reach you? I want to make sure this gets routed to someone who can look at your full setup and give you proper guidance on resilience and monitoring for a healthcare environment. Regards, Chiranjeet Vapi Support