We have recently noticed that there is spike in Call pricing because the number of tokens consumed per call are high. On further investigation, we noticed that for every tool call, instead of only sending tool call related parameters, VAPI is sending everything related toc all including the whole prompt, call details, assistant configuration etc. This is inflating the token count unnecessarily. Please advise if there is any option to turn off such additions...

Thank you for sharing the details. To help investigate further, could you specify whether you are using any custom tool integrations (such as MCP, Make, or Zapier), or are you seeing this behavior with Vapi's default/built-in tools as well? This will help determine if the payload size issue is related to a specific integration or a general platform behavior. Sources: - [Vapi Tools Documentation](https://docs.vapi.ai/tools) - [Vapi MCP Tools Documentation](https://docs.vapi.ai/tools/mcp)

I noticed the same, Token usage explodes.