You are rate-limited Rate-limited or API key is invalid

[JerryWn12]

using reverse proxy
proxy: https://chatgpt.pawan.krd/api/completions
model: text-davinci-002-render
api_key: https://chat.openai.com/api/auth/session's accessToken

[JerryWn12]

fixed

[PwLDev]

How did you fixed it? I'm having the same problem

[waylaidwanderer]

I heard it might be a bug with Pawan's API server. Use a free account with the text-davinci-002-render model.

[kilingzhang]

+1

[AutumnWhj]

How did you fixed it? I'm having the same problem

[JerryWn12]

I didn't do anything it just work automaticly🥲sorry just see the message
@PwLDev @kilingzhang @AutumnWhj

[AutumnWhj]

can you share your config ? 👁

[JerryWn12]

module.exports = {
    openaiApiKey: process.env.OPENAI_API_KEY || 'access_token_got_from_https://chat.openai.com/api/auth/session',
    chatGptClient: {
        reverseProxyUrl: 'https://chatgpt.pawan.krd/api/completions',
        modelOptions: {
            model: 'text-chat-davinci-002',
        },
        debug: false,
    },
    cacheOptions: {},
    bingAiClient: {
        userToken: '',
        debug: false,
    },
    apiOptions: {
        port: process.env.API_PORT || 3000,
        host: process.env.API_HOST || 'localhost',
        debug: false,
    },
    storageFilePath: process.env.STORAGE_FILE_PATH || './cache.json',
};

here is my config @AutumnWhj

[waylaidwanderer]

I see you set the model to text-chat-davinci-002. I think Pawan's API sets it to the text-davinci-002-render model if you don't give it a correct model name (like in your case), so keep that in mind.

[JerryWn12]

Thanks i will check my available model

[kilingzhang]

Still no effect ಥ_ಥ @AutumnWhj Does it work?

[waylaidwanderer]

Trying @acheong08's API server instead: https://chat.duti.tech/completions

Let me know if it works for you guys.

Edit: currently, his API endpoint throws an error due to the model parameter being set (his server doesn't accept extra parameters). He'll be updating it to allow setting the model, but this won't work until then.

[kilingzhang]

{"message":"Invalid request"}

[waylaidwanderer]

I'll try to deploy my own API server tomorrow as well btw

[waylaidwanderer]

For now, if you're having this issue, make sure to use a non-Plus account.

[kilingzhang]

ummm. Ok, I'll change to a non-plus account

[kilingzhang]

Still no effect ಥ_ಥ. It seems that I can only wait for the server deployed by you tomorrow to check again.

[JerryWn12]

Hey guys please notice that if you are using the reverse proxy, the openaiApiKey in settings.js should be the accessToken copied from https://chat.openai.com/api/auth/session but NOT the https://platform.openai.com/account/api-keys
@PwLDev @kilingzhang @AutumnWhj

[JerryWn12]

it seems like they patched this, the website no longer return the accessToken
*my mistake, i haven't login to chat.openai.com, after that the website returns accessToken again

[kilingzhang]

yes i confirm access_token not apikey ಥ_ಥ