Vapi is the Voice AI platform for developers.

Vapi AI

Hi, we are experiencing high latency on each turn of our calls. We've tried switching TTS and STT providers but it doesn't seem to help. Most calls get silence-hang-up.

<@1231163277078433874>
I'm sorry to hear that you're experiencing high latency issues. Latency can be affected by various factors, including network conditions, server processing time, and the efficiency of the API's infrastructure. To address this issue, we can look into optimizing the network path, reviewing the server's performance, and ensuring that the VAPI is configured correctly for low-latency operations.

Could you provide me with more details about your current setup? Specifically, information about the regions you are operating in, the average call duration before a hang-up occurs, and any error logs or metrics you have would be helpful. This will allow us to diagnose the problem more accurately and suggest appropriate solutions.

You can continue this conversation by mentioning me in the message.

example callIds:
f91c4c16-0096-4dd1-b4e0-2a66f2ae3b39
46340ade-0e24-4f81-b4c4-721c45e69a3e

After testing with different tts and stt providers it seems to be the LLM gpt-4o-mini - latency back to normal after changing back to gpt-4o - will continue to monitor

Alright, if you still get any latency issue can you let me know? <@1231163277078433874>

It seems the latency was only with 4o-mini - I don't think you currently support it officially

I can see directly calling OpenAi API with 4o-mini right now I'm getting >10s latency

We do support it, officially. But, let me check it from my side as well.

Oh ok great. That's good news! It seems like my issue is definitely related to OpenAI. Thanks for your help
https://cdn.discordapp.com/attachments/1266345837362151455/1266417211744976967/Screenshot_2024-07-26_at_17.29.33.png?ex=66a51270&is=66a3c0f0&hm=a6d1a063752f9d3b2eb0ae75ffd972b05db020e46f27eacce1c267eaf5c7b730&