Chrome Extension with Chat GPT-3.5 - "you must provide a model parameter"

Question

I am making a chrome extension that uses Chat GPT 3.5 and have coded a simple prompt to send to the API using openai api and returns a value in the console.

I have my code below and keep getting this error...

error: 
    code: null
    message: "you must provide a model parameter"
    param: null
    type: "invalid_request_error"

event though i have a model parameter.

// Define the API key
const API_KEY = "API KEY";

// Define the endpoint URL
const endpointUrl = "https://api.openai.com/v1/chat/completions";

// Define the headers
const headers = {
  "Content-Type": "application/json",
  Authorization: `Bearer ${API_KEY}`,
};

// Define the maximum number of completions to return
const maxCompletions = 1;

// Define the prompt to send to the API
const prompt = {
   model: "gpt-3.5-turbo",
   prompt: "Hello, world!",
   temperature: 0.5,
};

// Send a POST request to the endpoint with the prompt and headers
fetch(endpointUrl, {
  method: "POST",
  headers,
  body: JSON.stringify({
    prompt,
    max_completions: maxCompletions,
}),
})
  .then((response) => response.json())
  .then((data) => {
    // Log the response data to the console
    console.log(data);
  })
  .catch((error) => {
    console.error(error);
  });

Answer 1

I used your code and experienced the same error. I investigated the network request and saw that the payload was malformed:

{
   prompt: {
      model: "gpt-3.5-turbo", 
      prompt: "Hello, world!", 
      temperature: 0.5}, 
   max_completions: 1
}

So, as far as ChatGPT's api is concerned, you're only sending prompt and max_completions. The reason you request is formed this way is because you're passing an object filled with other objects into JSON.stringify().

Also, I'm not sure where you're getting the max_completions property from, as it is not in the API doc, so I left that out. Here is the change you need to make:

// in your fetch call:
fetch(endpointUrl, {
  method: "POST",
  headers,
  body: JSON.stringify(prompt), // prompt is an object, so no need to wrap it in another object.
}).then...

Another issue is that you're calling the create chat completion endpoint, but the properties you are sending are not correct. You are required to send:

• model: string
• messages: [{role: string, content: string}]

So, you would also need to make an edit here:

// in your prompt variable:
const prompt = {
   model: "gpt-3.5-turbo",
   messages: [{role: "user", content: "Hello, World!"}], // change prompt to messages array
   temperature: 0.5,
};

Cheers!