Hi, The way VAPI handles TTS from minimax will hit the rate limit on almost all their plans within a very short time. I have been testing MiniMax with VAPI, and I can’t even get a call to last a few seconds without hitting the rate limit. The issue seems to be that VAPI is constantly hitting minimax for fragments of a sentence, and instead of just sending the entire sentence and streaming the result, it splits up the sentence into multiple parts that will then be streamed? Can this be changed in anyway? As it stands, it does 10+ requests in <4 seconds of call time, which is a little aggressive imo.

To help reduce MiniMax TTS rate limiting, enable voice response caching in your configuration. This will cache repeated phrases and minimize duplicate requests:

json
{
  "voice": {
    "provider": "minimax",
    "cachingEnabled": true
  }
}

Set

cachingEnabled

to

true

in your MiniMax voice config to reduce API calls for repeated text ([documentation](https://docs.vapi.ai/changelog/2025/9/11)). Source: - [Voice Enhancements & Minimax Improvements](https://docs.vapi.ai/changelog/2025/9/11)

This doesn’t solve the issue

you can also try disabling chunking completely:

{
  "voice": {
    "provider": "minimax",
    "cachingEnabled": true,
    "chunkPlan": {
      "enabled": false
    }
  }
}

disabling or enabling chunking doesnt seem to do anything

if you are using your own API key for minimax, we suggest removing it. VAPI's default API key for minimax has a higher rate limit