Vapi is the Voice AI platform for developers.

Vapi AI

*At least* with custom LLM the first message can be interrupted regardless of this configuration option.

It appears that setting `firstMessageInterruptionsEnabled: false` should prevent interruptions of the first message, even when using `assistant-speaks-first-with-model-generated-message`. However, based on your experience, this may not be working as intended. As a workaround, try explicitly setting both options in your assistant configuration:

```json
{
  "firstMessageMode": "assistant-speaks-first-with-model-generated-message",
  "firstMessageInterruptionsEnabled": false
}
```

If the issue persists, it may be a bug—consider reporting it to Vapi support with your configuration details for further assistance.  
**Reference:** [Vapi Assistant API Reference](https://docs.vapi.ai/api-reference/assistants/create-assistant)

Source:
- [Vapi Assistant API Reference](https://docs.vapi.ai/api-reference/assistants/create-assistant)

Could you please provide some of the affected call IDs?

Hello, here is a call I just made to demonstrate the issue: 019d4ee6-c7f1-7117-85e7-23721491424a

The assistant configuration has `firstMessageInterruptionsEnabled: false` explicitly. I'm using custom-llm(BYO server) and the first message is generated. I can say anything to interrupt it and on this call I said "hello".

Expectation is that the message is uninterruptable and played in its entirety.

We're investigating this and will soon get back to you.