Could I make the ai sdk work through my local proxy?

[Svring]

My network is under censorship, so I have a proxy software that uses the port 33210 to handle all my requests(or part of them?) to guarantee network access, but in the ai sdk's case, all the requests to 'https://api.openai.com/v1/chat/completions' turned out to report timeout, while the requests sent from my browser and command line works without any problem.
The example code I used looks like this:

import { openai } from '@ai-sdk/openai';
import { generateText } from 'ai';
import { NextResponse } from 'next/server';

// Allow streaming responses up to 30 seconds
export const maxDuration = 30;

export async function POST(req: Request) {
  try {
    const { messages } = await req.json();
    
    const result = await generateText({
      model: openai('gpt-3.5-turbo'),
      messages,
    });
    
    return NextResponse.json({ text: result.text });
  } catch (error) {
    return NextResponse.json({ error: 'Request failed' }, { status: 500 });
  }
}

I'd like to ask if there is a way to let the request work through my local proxy, or if there is a way to monitor the request sent by my next.js project, so that I could find out a way to fix this problem.
Thanks for your attention.

[lgrammel]

You can configure the baseUrl, headers, and even provide a custom fetch function if needed: https://sdk.vercel.ai/providers/ai-sdk-providers/openai#provider-instance

[dev-bazz]

would be nice to see a code sample