Deleted old and installed that latest version, but this happens with Drafts:
Chat failed with HTTP status 429.
Body:
{
"error": {
"message": "You exceeded your current quota, please check your plan and billing details.",
"type": "insufficient_quota",
"param": null,
"code": null
}
}
There you go, a new version that shows tokens usage and cost, with an option in template tags to enable it or not.
I didnāt notice your previous reply about the app.displayInfoMessage method. I thought the only way to display some information was the alert function, which is why I believed it would be intrusive. Now we are on the same page, thanks again for your guidance.
Following the āUsageā link OpenAI API provided by @agiletortoise the page shows my āgrantā expired on 1 April. On the Pricing page, it says, āStart for free: Start experimenting with $5 in free credit that can be used during your first 3 months.ā I assume free use of the API is restricted by time (3 months) and/or credit ($5). There is also a note that the use of ChatGPT (web) and the API are priced separately. So buying the Premium plan wonāt allow the use of the API.
Hello! thanks for making this excellent action!
Is there a way to use this with the gpt-4 model?
I tired changing the Template Tag to gpt-4, but this told me that the model does not exist.
Any ideas how I can make this work?
Thanks!
OMG - I am so stupid! Sorry for bothering you. I somehow thought that because I am paying for API usage, that I automatically have access to GPT-4.
Thank you for your quick reply!
Hello! I have a seperate API key with access to GPT-4 now.
I would like to set up two separate ChatGPT Conversation Actions, one for GPT-3.5-turbo, and one for GPT-4, using credentials for multiple accounts. Could this work using the āCredential Identifier field" method? And if yes, where and how would I need to insert the āCredential Identifierā in the script? Thank you very much in advance and I hope this time around my question is not as dumb as last time
This action talks directly to OpenAI and does not use Drafts built-in OpenAI scripting, it is storing itās own custom API key using the credential system, but not sharing the āOpenAIā. To have a copy of it use a different API, you would only need to alter the first line to use a different initializer for the Credential object with a different name - this change at the beginning of the script should do it:
// change this line...
const credential = Credential.create('ChatGPT Conversation', "Credentials for ChatGPT Conversation");
// to this...
const credential = Credential.create('ChatGPT Conversation (GPT-4)', "Credentials for ChatGPT Conversation");
Note the first string parameter passed it changed. What you change it to does not really matter - but thatās how it will appear in Settings > Credentials if you ever need to reset it.
Hi folks! Hope you are fine! I have got a problem with the ChatGPT Conversation Action and the API call! It stops working with this message: Chat failed with HTTP status -1.
Body:
undefined
It pops up when running this action right after prompting and with context prompting. Also resetted the API key. Changed the network (wifi / mobile /). No success.
Maybe someone can give me a hint or advice? Greetings!
requests have been timing out for a few hours, from Mac and from iOS. openAI seems to be doing just fine on their website. Did they change something with how API requests work?
Iāve installed this plug-in and added my API key, switched the model back to GPTChat-3.5 (since Iām using the free version) and I get the following error:
Chat failed with HTTP status 404.
Body:
{
āerrorā: {
āmessageā: āThe model gpt-3.5 does not exist or you do not have access to it.ā,
ātypeā: āinvalid_request_errorā,
āparamā: null,
ācodeā: āmodel_not_foundā
}
}
I changed to 3.5 by editing the action and then edited the template and changed the gpt-4 to gpt-3.5. Any idea what Iām doing wrong?
Ah, ok that made a difference, but now Iām getting the 429 āinsufficient_quotaā error. I thought version 3.5 was free to use now?
Maybe itās only free if you use it from their website and not an API?