Feature Requests

NVIDIA just released PersonaPlex, an open-source AI voice model that can listen and speak at the same time with almost zero latency.

It seems like NONE of the AI Voice Agent companies seem to have a Custom Client Billing feature where we can auto-bill our clients based on their call minute usage/overage/pro-rate or a flat monthly charge to them, etc. If this is at all possible to do, it would be a feature that only you have - a great selling point for Vapi (and make it easier for us, lol). Thanks, T

The idea is to be able to use voice agents but in WhatsApp as texting also.

Issue: When using Vapi's built-in Langfuse integration, tool call messages are logged with camelCase field names instead of snake_case, making them incompatible with the OpenAI API format. Current behavior: Messages logged as: toolCalls , toolCallId OpenAI API expects: tool_calls , tool_call_id Impact: Langfuse Playground cannot replay conversations containing tool calls because it receives a 400 error: "Invalid parameter: messages with role 'tool' must be a response to a preceding message with 'tool_calls'." Steps to reproduce: Enable Langfuse integration in Vapi (Settings → Integrations) Create an assistant that uses function/tool calls Make a call that triggers a tool View the trace in Langfuse Select a Generation after the tool call Click "Playground" button Click "Submit" Observe 400 error Expected behavior: Tool call messages should use snake_case ( tool_calls , tool_call_id ) to match OpenAI API spec, allowing Playground to function correctly.

It would be nice if we could see the reason why the assistant request failed directly from the calls logs. Currently, the way I'm debugging it, is by logging the request and trying to create an assistant (which gives me the error).

We can set a custom name for our own reference, but we can't list all with that reference. That's would be wonderful if that worked. That way I make multiple calls/retries for the same purpose and list all attempts.

There’s a new model from inworld that i can’t select as the list is limited to older models. Can you please add inworld-tts-2 to the model picker. Thanks!

please add openai/gpt-chat-latest (GPT-5.5 Instant)

Request WhatsApp capability for text chat with Vapi in addition to SMS inbound. WhatsApp can be logged for compliance purposes in regulated industries whereas SMS cannot. Enabling WhatsApp would be material for some client types.