Fallback models from different providers

moderate-tomato · 2025-05-07T09:17:13.742Z

Is there a possibility to configure a fallback model from google (e.g. gemini-flash-2) when the default is set to openai. Would be amazing incase one provider fails completely

V

Vapi Support BotAPP•5/7/25, 9:18 AM

Yes, you can configure a Google model (gemini-flash-2) as a fallback when your default is set to OpenAI in Vapi. According to Vapi's documentation, you can add this to your assistant's configuration:

{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "fallbackModels": [
      {
        "provider": "google",
        "model": "gemini-flash-2"
      }
    ]
  }
}

{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "fallbackModels": [
      {
        "provider": "google",
        "model": "gemini-flash-2"
      }
    ]
  }
}

{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "fallbackModels": [
      {
        "provider": "google",
        "model": "gemini-flash-2"
      }
    ]
  }
}

{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "fallbackModels": [
      {
        "provider": "google",
        "model": "gemini-flash-2"
      }
    ]
  }
}

This helps protect against provider-wide outages.

Sources:

⤵️ Mention @Vapi Support Bot for follow-ups 🐥

V

Vapi Support BotAPP•5/7/25, 9:18 AM

Yes, you can configure a Google model (gemini-flash-2) as a fallback when your default is set to OpenAI in Vapi. According to Vapi's documentation, you can add this to your assistant's configuration:

{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "fallbackModels": [
      {
        "provider": "google",
        "model": "gemini-flash-2"
      }
    ]
  }
}

{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "fallbackModels": [
      {
        "provider": "google",
        "model": "gemini-flash-2"
      }
    ]
  }
}

{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "fallbackModels": [
      {
        "provider": "google",
        "model": "gemini-flash-2"
      }
    ]
  }
}

{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "fallbackModels": [
      {
        "provider": "google",
        "model": "gemini-flash-2"
      }
    ]
  }
}

This helps protect against provider-wide outages.

Sources:

⤵️ Mention @Vapi Support Bot for follow-ups 🐥

E

extreme-purple•5/7/25, 9:45 AM

By setting up a custom LLM endpoint that tries OpenAI first and switches to Gemini Flash 2 if it fails, then linking that endpoint as a Custom LLM in your assistant settings.

Eextreme-purple By setting up a custom LLM endpoint that tries OpenAI first and switches to Gemi...

W

worthy-azure•5/7/25, 10:06 AM

Can i contact you please would love some help with current issues facing with Vapi

V

VapiAPP•5/10/25, 11:15 PM

Hey Bank, checking if this is resolved/solved for you?

Ccomprehensive-tomato Hey Bank, checking if this is resolved/solved for you?

M

moderate-tomatoOP•5/13/25, 9:09 AM

Hey Shubam, unfortnunately not. We dont want to switch to a custom LLM due to latency reasons, and i cant see a possibility to set a fallback model from a different provider. Would be amazing if this is possible

V

VapiAPP•5/14/25, 11:26 PM

Hey,

Could you please schedule a brief call at your convenience so we can discuss this matter in detail?

Kindly use the following link to book a suitable time: https://calendly.com/shubham-vapi/chat-30m.

S

scattered-teal•10/10/25, 4:36 PM

Hey all, we're facing the same issue.
We're using grok, and I guess that sometimes some error happen and the assistant calls OpenAI, but the model is very slow. I would like to at least select which model it is going to use on openai. I tried the Support bot solution, but I get 404 when I send the request, and I can't find anything about it on the docs.

V

VapiAPP•11/18/25, 4:32 AM

I sincerely apologize for the delayed response to your support request.
We implemented better tracking for support tickets to ensure timely responses going forward, but due to a technical limitation, this only applies to tickets created after this past week.
I'm closing this older ticket, but if your issue still needs attention, please:
Reply with 'REOPEN' or
Email support@vapi.ai with any updates
I'll personally ensure your new request gets prompt attention.
Thank you for your patience.
Best regards,
Vapi Support Team

Fallback models from different providers

Similar Threads

Similar Threads

Similar Threads