API Rate Limit issue

[RichGapple]

Duplicates

• I have searched the existing issues

Steps to reproduce 🕹

Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...
Error: API Rate Limit Reached. Waiting 10 seconds...

This is the only error im getting and I have run python scripts/main.py --gpt3only

Current behavior 😯

No response

Expected behavior 🤔

No response

Your prompt 📝

# Paste your prompt here

[tuapuikia]

Open API chatgpt is not a free service. You can resolve this by adding credit card to your openai account.

[AramisValdes]

Having the same issue, i have premium GPT account.

[Slamsneider]

"premium GPT account" is not enough. You need to throw your creditcard here: https://platform.openai.com/account/billing/overview and make a $ account for tokens.

Then you can also use the ($) playground. https://platform.openai.com/playground

Remember to set a soft and hard limit or you can loose control of your money spending very quickly.

[disneytoy]

Same issue. I just paid for GPT Plus $20 a month. So that isn't enough to use the API?

[tuapuikia]

Chatgpt plus is for web chat, API have different billing.

[marcocastignoli]

I know some libraries use the web token to interact with chatgpt instead of using OpenAI API token. Are there mandatory features in the api that prevent you allowing users to verify themself through the web token?

If not, I'd be available to integrate the web token.

[tuapuikia]

I know some libraries use the web token to interact with chatgpt instead of using OpenAI API token. Are there mandatory features in the api that prevent you allowing users to verify themself through the web token?

If not, I'd be available to integrate the web token.

Is it possible to specify GPT4 model when using the web token method?

[marcocastignoli]

I just checked a request to the web api.

{"action":"next","messages":[{"id":"6bee6a9c-8211-4597-a602-b695a2787615","author":{"role":"user"},"content":{"content_type":"text","parts":["hello"]}}],"parent_message_id":"9fde1d70-23d7-4e21-9826-5d67ae88f397","model":"gpt-4","timezone_offset_min":-120}

Seems like you can specify the model.

[JuroOravec]

Duplicate of #1189. @patrickelectric would you close this one issue if it indeed is the same thing?

(I had this issue too - It's that we haven't opted into "paid account" on OpenAI's billing page)

I'm making an MR as we speak to update the README section on OpenAI's API key, to make this clear.