----------------------------- **While I generate an answer, here are some sources that might be helpful**: * [Frequently Asked Questions]() * [Introduction]() * [Custom LLMs]() * [Create Assistant]() -----------------------------

### Answer Yes, it's possible to integrate your custom LLM and knowledge base with Vapi. You can utilize Vapi's support for Custom LLMs, allowing you to use any OpenAI-compatible endpoint, including your own server. Here’s how you can set it up: 1. Using an LLM Provider: If your custom LLM is hosted with an LLM provider compatible with OpenAI's API, you can POST your API key via the

/credential

endpoint and specify your custom model in your API requests [(1)]() . 2. Using Your Server: For LLMs hosted on your server, you can create an endpoint compatible with the OpenAI Client. Your server can then interact with your knowledge base and vector store as needed. To implement this, post your server’s API key and URL via the

/credential

endpoint and specify these in your assistant creation with the

custom-llm

model provider. This setup allows Vapi to provide transcripts to your specified endpoint, which can then utilize your knowledge base for responses. For more detailed instructions on setting up your custom LLM with Vapi, refer to the Custom LLMs documentation.

It is kinda similar to this https://github.com/VapiAI/server-side-example-serverless-vercel