karthink / gptel

A simple LLM client for Emacs
GNU General Public License v3.0
1.06k stars 116 forks source link

ChatGPT error: (HTTP/2 429) You exceeded your current quota, please check your plan and billing details. #70

Closed AbeEtan closed 1 year ago

AbeEtan commented 1 year ago

Hello everyone,

I am getting the following error

ChatGPT error: (HTTP/2 429) You exceeded your current quota, please check your plan and billing details.

I can use ChatGPT through the webrowser just fine.

Note: I haven't been able to run anything locally yet so I didn't even use any quota.

Please advise

karthink commented 1 year ago

@AbeEtan I don't think there's a free quota any more, have you added your billing details to your account at openai.com?

Kabouik commented 1 year ago

I haven't added any billing data to my account, and would prefer not to if I can avoid it, but like @AbeEtan, I can use ChatGPT 3.5 just fine in the web app by OpenAI, and get this same error with gptel.

Does using anything else than the web app, i.e., relying on the API key, necessarily involves a cost? The Usage page in OpenAI account settings does suggest that everyone is given $18 of free use (I'm at 2$), but it's not clear whether that it's every month or expired on May 1st. That's a shame if the API is now paid, my use would be the same as on the web app but it is just really slow and CPU intensive on my end.

Kabouik commented 1 year ago

I can confirm that setting up an account does solve this issue. Too bad, I'd love to use gptel more, but that makes it have a cost while the web app is still free (and inconvenient).

PalaceChan commented 1 year ago

yes i had to do that too after my free trial. only the (garbage) website api is free. but pay per use is extremely cheap (especially with gpt 3.5) and this package is too good not to be leveraging it for the fractions of a cent it ends up costing

karthink commented 1 year ago

yes i had to do that too after my free trial. only the (garbage) website api is free. but pay per use is extremely cheap (especially with gpt 3.5) and this package is too good not to be leveraging it for the fractions of a cent it ends up costing

I have been using the ChatGPT 3.5 API for a few months now and am yet to break 2 cents on the bill. However I share the concern about trusting OpenAI with your credit card details and about supplying them with even more data. When using local or open-source LLMs becomes easier I'll look into adding support for them to gptel. Right now they're (a) not as good, and (b) quite inconvenient to set up and use.