Difference between newer and older model endpoints?

API
1

Hi all I’m working with an older codebase that uses some custom code to interact with the davinci model by providing the openai key in the headers, specifying a model and calling this endpoint: https://api.openai.com/v1/completions

I want to upgrade this to “text-davinci-002” with minimal changes to the existing code so I tried omitting the model parameter and calling this endpoint instead, as seems to be recommended in the latest docs https://api.openai.com/v1/engines/text-davinci-002/completions

This seems to give an invalid_request_error “message”: "You didn’t provide an API key.

However, if I do the same thing with the new davinci endpoint https://api.openai.com/v1/engines/davinci/completions

it works great!

I am a bit confused at why that is the case, has anyone else noticed this and any suggestions?

1 Like
2 
const completionEndpoint = "https://api.openai.com/v1/engines/text-davinci-002/completions";

const openaiHeader = {
  "Content-Type": "application/json",
  Authorization: "Bearer " + OPENAI_KEY,
};

const generateText = ({
  prompt,
  max_tokens = 1000,
  stop = ["END", "***", "---", "\n"],
}: {
  prompt: string;
  max_tokens?: number;
  stop?: string[];
  model?: string;
}): typeof SSE => {
  return new SSE(completionEndpoint, {
    headers: openaiHeader,
    method: "POST",
    payload: JSON.stringify({
      prompt: prompt,
      temperature: 0.75,
      top_p: 0.95,
      max_tokens: max_tokens,
      stream: true,
      stop: stop,
    }),
  });
};
1 Like