DALL-E 3 Rate Limit Discrepancy & Increase Request

API Bugs
, ,
1

I’m encountering a critical inconsistency between documented rate limits and actual API restrictions for DALL-E 3.

Issue:

Error code: 429 - {'error': {'code': 'rate_limit_exceeded', 'message': 'Rate limit exceeded for images per minute in organization {...}. Limit: 15/1min. Current: 16/1min. Please visit https://platform.openai.com/docs/guides/rate-limits to learn how to increase your rate limit.', 'param': None, 'type': 'requests'}}

This is preventing my application BulkImageGeneration [dot] com from supporting user demand for DALL-E 3 integration. My infrastructure currently handles 60,000+ images/minute via Replicate without issues, but users strongly prefer native OpenAI integration.

Request:

  1. Investigate/resolve the documentation vs. implementation mismatch.
  2. Increase my organization’s rate limit (org-2yXbmvRDAX7GllNx9m2hUV1S) to at least 7,500 images/minute as advertised.

Why this matters:

  • My users rely on high-volume generation for workflows like e-commerce, marketing, and AI prototyping.
  • Current limits make OpenAI’s API unusable for our use case despite clear product-market fit.

Thank you for your urgent attention to this matter

1 Like
2

Was just looking at the same thing. The OpenAi documentation says that Rate Limit for Tier 3 For Dalle3 is 5,000 img/min. My organization account page says that I am on Tier 3 and Dalle 3 image limit is 5,000 RPM 7 images per minute.

3

I think that the rate limit documentation page, the one where you can select any tier and see rates, is a bit borked up.

Here’s the image rate I recorded by actual tier reports six months ago.

    {
        "class": "dalle2",
        "id": [
            "dall-e-2"
        ],
        "rpm": [
            5,
            5,
            50,
            100,
            100,
            500
        ]
    },
    {
        "class": "dalle3",
        "id": [
            "dall-e-3"
        ],
        "rpm": [
            1,
            5,
            7,
            7,
            15,
            50
        ]
    },

That’s starting at tier “0”, through tier 5.

There are two different values now shown on your own organization’s limits page, a very high “RPM”, and a line “NN images per minute”.

Perhaps the large RPM are rates for API inputs to the image models, just in terms of allowed endpoint requests, and following that, another for successful image generations by model (in case you get denials from prompt or content)?

Rates seem to now be a bit higher on the highest tier 5.

image

4

Thank you for sharing your numbers. In this case the RPM does not make sense to me. Also the image generation rate is so low even for higher tier. Seems way too low for any professional use case I can come up with.