Bad Request 400 status code: "You must provide a model parameter" when attempting any openAI .call method

[snapdeus]

Greetings,

When using Node 18.1.0, or lower, on both Ubuntu 20.0.4 and Windows 10:

Attempting to use the method .call() results in a 400 status code.

Here is my code:

import { ChatOpenAI } from "langchain/chat_models"
import { HumanChatMessage } from 'langchain/schema'
import dotenv from 'dotenv'
dotenv.config()

const chat = new ChatOpenAI({
    openAIApiKey: process.env.OPENAI_API_KEY,
    temperature: 0.9

})
async function getChat() {
    const response = await chat.call([
        new HumanChatMessage(
            "Translate this sentence from English to French. I love programming."
        ),
    ]);


}

getChat()

Things I have tried:

• Instead of passing the openAIApiKey as a parameter with .call(), I instead ran package.json script: npm run dev which was defined as node -r dotenv/config index.js
• Downgrading to Node 16.8.0 and setting NODE_ENV to experimental
• Downgrading the version of langchain to 0.0.22

TEMPORARY SOLUTION

After changing my Node version to 18.2.0, the error does NOT happen!

What I suspect may be the cause:

Something with AxiosFetchAdapter...that's the only thing I can think of!

[nfcampos]

Hi

In general it's not recommended to stick with old minor versions of Node. We only support running Langchain on the most recent versions of Node.js 16.x, 18.x, 19.x

[jacobcwright]

Still getting this error with node v18.16.0

[nfcampos]

Duplicate of #1010

[Developergovind]

same error