Assistant API: usage of Knowledge Retrieval tool in the Run is blocking Function calling

Openai claims that " Assistants can access multiple tools in parallel"

I am using both Knowledge Retrieval and Function calling in my Assistant.
And when the model has to reply using knowledge from the the Knowledge file (god knows how it decides), the run, which I create, is not interrupted if the “requires_action” status (it should on every message according to the instruction).

When the model is provided by an input message, which couldn’t be addressed in the Knowledge file - the Function calling works fine.

Can Knowledge Retrieval actually intercept the Function calling?

I have been able to use function calling with knowledge files using GPT-4-1106-preview model. With GPT-3.5-turbo-1106, I have not got it working. Also, due to price with GPT-4 and the quality of knowledge retrieval did not yet meet my bar to make more use of it.

I am also using it with GPT4-1106-preview.
But had your function been called in the same Run, in which the assistant message is generated based on file knowledge?