Problem with creating GPT4o-Vision Batch (Enqueued Tokens Limit)


I’m trying to create a Batch Job with GPT-4 Vision. However, even when the batch only has a few lines, I get the following error:

“Enqueued token limit reached for gpt-4o in organization org-… . Limit: 90,000 enqueued tokens. Please try again once some in-progress batches have been completed.”

There are no batches in progress, and every batch size I’ve tried has failed.

If someone knows how to fix this, I would appreciate it.

Thanks in advance!

Welcome to the Forum!

If this batch queue limit is too low for you, then you need to consider moving to a higher usage Tier. Currently, it appears that you are in Tier 1. In order to move to a higher level, which then comes with a higher limit, you need to add more funds to your account. You can find the overview of eligibility criteria for the different tiers in the screenshot and link below.


Thanks for the quick answer.

I understand that my limits are not that high, but I am not exhausting my limits even in the slightest. I even tried it with a batch that only needs a few tokens, but I received the same error.

Am I misunderstanding something?

How large are your requests? What’s the size of your input/output tokens per request?

Or let me ask differently: have you tried to run just a single regular API call with one of your requests to get a better breakdown of your token usage? That would help to identify the potential root causes.

1 Like

Thanks again.

I was able to create a batch small enough to run. The only thing I don’t understand is that I was able to run a batch with more than the token limit with GPT-4-turbo just fine a month ago. Is this limit new?

No, the limits are not new. For gpt-4-turbo the same limit of 90,000 tokens applies under Tier 1. So not sure why this was possible :thinking: