extra_query parameter in OpenAI.batches.list

[Cybernetic-Ransomware]

I am trying to filter batches on the server side using the extra_query parameter like this:
extra_query={'status': 'completed', 'created_before': '2024-08-26T00:00:00Z'}

Unfortunately, it's not working. I also couldn't find any examples. Is the extra_query parameter currently functional?

[Hikari-Tsai]

extra_query will append arbitrary query parameters to the HTTP request, but it does not make the server support filters that the endpoint does not define.

For client.batches.list(), the current API reference only documents after and limit as query parameters:
https://developers.openai.com/api/reference/resources/batches/methods/list/

So parameters like status or created_before are not server-side filters for this endpoint. You need to page through the batches and filter client-side, for example:

from datetime import datetime, timezone

cutoff = datetime.fromisoformat("2024-08-26T00:00:00+00:00")

for batch in client.batches.list(limit=100):
    created_at = datetime.fromtimestamp(batch.created_at, tz=timezone.utc)
    if batch.status == "completed" and created_at < cutoff:
        print(batch.id)

If you have many batches, use pagination (limit / after) or the SDK's auto-pagination pattern and stop once your ordering assumptions make it safe.

[Cybernetic-Ransomware]

Thanks, that clarifies it.

So extra_query only appends arbitrary query parameters to the request, but batches.list() currently supports only after and limit server-side. Filtering by status or created_at has to be done client-side.

That works, but it means clients with many historical batches must paginate through potentially all batches just to find completed or older ones. Server-side filters for status and creation timestamps would be very useful here.