Inconsistant Top K limit in Gemini API and Vertex AI

Gemini API
, ,
1

I am migrating a project from Gemini API to Vertex AI.
When using Gemini API I’d set the top_k in generation_config to 64.
However when I try the same generation with Vertex AI, it throws this error:

{
  "error": {
    "code": 400,
    "message": "Unable to submit request because it has a topK value of 64 but the supported range is from 1 (inclusive) to 41 (exclusive). Update the value and try again.",
    "status": "INVALID_ARGUMENT"
  }
}
3

A previous discussion confirms that vertex has a maximum top_k limit of 41:

:link: Discussion on Google AI Forum

However, this limit is not mentioned in the official documentation:

:link: Google Cloud Documentation