As of 2 days ago, running the Batch API using the gpt-5 model is producing the following error:
The model
gpt-5-2025-08-07-batchdoes not exist or you do not have access to it.
Our Batch API is using the Responses API endpoint, so haven’t checked if it’s the same with other endpoints. Everything was running fine prior to this weekend with Batch API and gpt-5.
Anyone else facing this issue suddenly? Any ideas for resolution would be appreciated - otherwise, we might just have to wait.
If the internal model name with “batch” is being exposed, it is failing in a way others have experienced.
Solution to try: create a new project for the batch requests and batch file retrieval, and an API key without setting any limits other than default “allow”.
@edwinarbus Is this a known issue that’s being worked on? Just checked this morning and Batch API still doesn’t work with gpt-5.
Based on our batch records, everything worked fine until Sunday morning (April 10), then we started having this issue Monday morning (April 11).
I’m getting the same issue starting at the same time, and it appears others are hitting the same thing with batching in another thread (it won’t let me post links). Is anyone currently able to use the batch API?
I am getting the same issue earlier today. GPT-5 batch API cannot be called through responses endpoint, though it works on the chat.completions endpoint.
Hopefully it can be solved very soon!
Still getting this issue. I tried creating a new project and am getting the same error. I just uploaded the file in the OpenAI platform and initiated the batch process against it.
Error:
The model gpt-5-2025-08-07-batch does not exist or you do not have access to it.
Thanks for the report! We’ve identified the issue and are currently working on a fix. Will post again once it’s resolved.
This is resolved and verified now, thanks again for the report and sorry for the trouble!
The problem persists for me for certain projects but not others.