The documentation used to say
The moderation endpoint is free to use when monitoring the inputs and outputs of OpenAI APIs. We currently disallow other use cases.
Now it says
The moderation endpoint is free to use for most developers.
https://platform.openai.com/docs/guides/moderation/overview
Does this mean now I can use the Open AI Moderation API against non-Open AI generated content and prompts?
Sounds more like “is not free to use for…” (insert accounts who must pay soon).
Interesting. Good catch.
I wonder if it will be free for OAI users/uses and a fee for others?
Could be a tidy little revenue stream for them, I imagine.
This was changed a little over two weeks ago,
The full list of changes for this help article is,
The moderations endpoint is a tool you can use to check whether
content complies with OpenAI’s usage policiestext is potentially harmful. Developers canthuse it to identify content that ourusage policies prohibitsmight be harmful and take action, for instance by filtering it.
The moderation endpoint is free to use
when monitoring the inputs and outputs of OpenAI APIs. We currently disallow other use cases. Accuracy may be lower on longer pieces of textfor most developers.
Below is an example output of the endpoint. It returns the following fields:
- flagged: Set to true if the model classifies the content as
violating OpenAI’s usage policiespotentially harmful, false otherwise.- categories: Contains a dictionary of per-category
binary usage policiesviolation flags. For each category, the value is true if the model flags the corresponding category as violated, false otherwise.
OpenAI willWe plan to continuously upgrade the moderation endpoint’s underlying model. Therefore, custom policies that rely oncategory_scorescategory_scoresmay need recalibration over time.
My read on this is OpenAI may be looking to expand the use and usefulness of the moderation endpoint beyond filtering what is sent to the models. I imagine use for the original purpose of preventing bad inputs from making it to the model will continue to be free, limited light use for other purposes might be free, and use at scale may require some sort of paid contract.
But, absent an official announcement from OpenAI, it would likely be most prudent to continue to use the moderations endpoint as originally intended.
Yeah, it could be a nice “little” revenue stream for them.
Solid advice, @elm!
My read of this is that OpenAI phrased any commitment on the side of OpenAI out of it.
Read:
You cannot use the moderations endpoint to determine whether content violates OpenAI’s usage policies or not. Use of the moderations endpoint does not indemnify you as a developer, and we may still close your account for submitting false negatives to the model.
OpenAI makes no commitment to upgrading the moderation endpoint, but it’s possible that it may be planned.
I think it would be great if they expanded the use of this model. I would love to use it in other applications that do not have to do with OpenAI.
From the FAQ:
The moderation endpoint is free to use when monitoring the inputs and outputs of OpenAI APIs. We currently disallow other use cases.
Looks like the article has been updated:
Yes, the Moderation endpoint is free for OpenAI API users, and usage of this tool doesn’t count towards your monthly usage limits.
https://help.openai.com/en/articles/4936833-is-the-moderation-endpoint-free-to-use
Now I’m more confused by OpenAI’s stance on this, hmm
I seem to remember them announcing that you could now use it with non-OpenAI content a few months ago.
I would read this like a sign in a store saying “the bathroom is for customers only.”
Maybe you bought something there last week, that doesn’t make you a customer today for the purpose of determining if you can use the bathroom.
Likewise, just because you are (even on an ongoing basis) a user of OpenAI’s API services, that doesn’t necessarily make you a “user” for the purpose of determining if you can use the moderation endpoint for free.
I imagine they’re generally pretty lax on this, but I think the takeaway should be that the moderation endpoint is free if you’re using it in conjunction with some other OpenAI API.