Skip to main content
Question

Rate limit reached for gpt-4 prevention

Hi everybody,

I’m making my first few zaps with Chat-GPT4 this week.

I keep getting a Zap alert about the rate limit reached for gpt-4.

My process:

  • Upload 20-40 files to google drive at the time
  • Before Chat GPT conversation, i put Zapier delay after queue with
    • unique title for every file
    • 0.5 minutes

After 4-5 files i get e429] Rate limit reached for gpt-4

Im at Usage tier 1 at Open ai playground
My max Tokens limit in zapier is set to 250, but i dont think it does anything, since it uses up to 10 000 tokens until it fails, so is a Open AI problem.

I can’t seem to figure this out. I feel like I’m missing something so simple

3 replies

SamB
Rank
Userlevel 7
Badge +11

Hello there @Marekk, welcome to the Community! 🎉

Are you uploading the files to ChatGPT? If so, then the errors might have related to this incident on OpenAI’s end: https://status.openai.com/incidents/vn8j63zg56hz

If not, are you able to replay the errored runs of the Zap successfully? The rate limit documentation here for tier 1 suggests that it would be limited to 500 requests per minute so 4 or 5 files a minute shouldn’t be hitting that. But if it’s possible for them to be replayed successfully later on maybe the delay needs increasing to spread out the requests more. 

Can you give that a try and let us know how it goes?

A

Hi,

I’m facing a similar problem with my zap and the rate limit. I started working with ChatGPT 4o recently and I am constantly hitting the rate limit with my zap right now. My Zap is designed this way:

  1. ChatGPT is supposed to memorize a text from a Google Drive File, so that the information is stored in the Conversation ID
  2. There is a Loop to repeat this step for every text that is added to my Google Sheet (The GoogleSheet is the trigger for the Zap)
  3. After the Loop, ChatGPT is supposed to make a summary of all texts it memorized

The problem I’m facing right now is that in the first step with ChatGPT memorizing the texts, it already reaches the rate limit for 4o. The texts are not that long, so the input is very similar each time and should not exceed the rate limit. I have the feeling the problem might be the conversation ID. But without using it, the whole step would not work as required. I have already tried to put in delays to prolong the time between the ChatGPT steps activating, but it does not always work.

Does somebody have an idea, how to work around this rate limit problem?

SamB
Rank
Userlevel 7
Badge +11

Sorry to hear you’re also running into these rate limiting errors as well @acDE.

I can see that you’ve also reached out to our support team and they agree that the issue is likely due to the amount of information being stored for the conversation ID being used. And they spotted that the file object (displays as “File Pdf: (Exists but not shown)”) had been mapped to the Message field on the Conversation With Assistant action which might be causing some trouble.

I wonder if the involvement of the files might be causing it to hit these limit errors 🤔 Perhaps instead of storing the conversation text in OpenAI directly maybe you could keep a record of them in a Zapier Table? Suggesting Zapier Tables as it wouldn’t count towards your task usage. For that approach you could use Find Record (Zapier Tables) action to obtain the necessary information that ChatGPT needs to reference and map it to the Message field. And you’d use an Update Record in the loop to add more information to the existing record in the Zapier table. 

Do you think that might work? If you give that approach a try please keep us updated on how you get on, would love to know whether or not that helps to get past those rate limit errors! 🤞

Reply


Terms & Conditions