Limits for the generation of journalistic news of a web platform? Info on number of requests and subscription plans

I’m new to the world of OpenAI and ChatGPT. I would like to create a private website, reserved for the editors of my online newspaper. It will be a private platform with the purpose of generating news articles to be published in my newspaper. I have doubts about the limitations of requests in relation to hours or in a day.

I should advertise about 30 to 50 news articles per hour (1 hour). Obviously, this pace will not be the same for all hours because there will also be hours when I can only publish 10 news articles in an hour (1 hour).

Let’s take, for example, the maximum limit of news articles I should publish in 1 hour, which will be between 30 and 50 news articles. Consequently, there should be between 30 and 50 requests per hour.

Input tokens should be around 300-400 and output tokens should be around 300-400 too.

In the private news generation platform, only 4 or 5 editors will have access to generate the news and they will divide the workload. Each user will log in at the same time, and each will generate news with a variable number of requests (for example, one user will generate 10 news articles, another user will generate 20, another user will generate 5) until a total of generally between 30 to 50 news articles in 1 hour.

I thought the limits depended only on the number of API tokens, which increases the total budget cost. So I thought the solution was to have a substantial budget for the API. But I also became aware of the limits on the number oBut reading on the web I also learned about the limits of the requests that can be made in 1 hour or in a day. Having presented you with the characteristics of my project, can anyone tell me if it will be possible to generate between 30 and 50 news items in 1 hour?

What is the limit of news articles that I can generate in 1 hour? Can you explain how it works in relation to my project? Do I need to subscribe to Plus or Team, or will they not be sufficient?

Is there anyone so kind who can explain to me in a simple but detailed way considering that I’m new and would like to understand how it works? Thank you

I think this is feasible, although I’m not entirely sure of the project’s intent. You could switch between different models or assign individual API keys to each editor, which they would use to generate completions. This approach could potentially increase the total available quota by utilizing multiple keys. While some API models have limitations and varying performance levels, you could use one model for content discovery, another for organization, and a third for writing articles. This shouldn’t be a decisive factor for the project’s success.

@williampferebee Thanks, but there are still some unclear things:

1. Why should we assign individual API keys to each editor? What difference does it make if I use a single $1000 API key (to give a large example) and pass it to all the editors, or if I assign a different key to each editor? What is the advantage of assigning multiple keys to each editor?

2. I would like to know more about the number of possible request limits. I asked Chatgpt and he replied like this:

Can you confirm that these are the limits of the requests?

My idea wasn’t very fleshed out, and it appears that rate limits are tied to the account level and not the API key, however you could utilize different models in your workflow to limit the risk of hitting a rate limit. And you could use different API keys for different models to help you keep track of usage, though that is not strictly necessary since any API key will work for any model.