According to the documentation, the file_search tool of the Responses API admits several vector store ids:
vector_store_ids array Required
The IDs of the vector stores to search.
But adding more than one throws an error:
"Invalid 'tools[0].vector_store_ids': array too long.
Expected an array with maximum length 1, but got an array with length 2 instead."
Is this something to be fixed by Open AI?
It’s documented:
True. Not in the API reference, unfortunately. And both are contradicting.
Anyway, I hope they allow searching in more than one vector store, as it seems to be the plan.
It looks like they have increase the ability to query two vector stores now. But when I tried to add the third vector store, I receved the error message below. Well, two is still better than one, I hope they can increase the limit even further.
{
“error”: {
“message”: “Invalid input: maximum of 2 vector stores allowed.”,
“type”: “invalid_request_error”,
“param”: “tools”,
“code”: “integer_above_max_value”
}
}
Are there any news on this? I tested it in the playground with two VS, but it seems to know nothing about the content of the second VS. When can we expact to use two or more VS?
thanks
File in VS 1 says “We love dogs” , when i ask the question “what do we love, the answer is: we love dogs”
When i switch the VS (File in VS 2 says: “We love cats”) the answer is “we love cats”.
My expectation was, that it says “we love both”. It does, when i put both files in one VS.
So, it can only see the first VS, not the second.
I am finding the same problem. Was there a solution to this?
