Model not found gpt-4o-mini-search-preview (invalid_request_error) with latest version

[oboe-ori]

Description

In a new project, add to your dependencies:

"dependencies": {
    "@ai-sdk/openai": "^2.0.69",
    "ai": "^5.0.98"
  }

Make a simple generateText (or generateObject or streamObject) request. Get error from OpenAI. This worked in previous versions.

import { generateText } from "ai";
import { openai } from "@ai-sdk/openai";

async function main() {
  const { text } = await generateText({
    model: openai("gpt-4o-mini-search-preview"),
    prompt: "Write a short poem about TypeScript",
  });

  console.log(text);
}

main().catch(console.error);

Set OPENAI_API_KEY environment variable and run above code. You will get an error (all other models work):

  responseBody: '{\n' +
    '  "error": {\n' +
    '    "message": "Model not found gpt-4o-mini-search-preview",\n' +
    '    "type": "invalid_request_error",\n' +
    '    "param": "model",\n' +
    '    "code": null\n' +
    '  }\n' +
    '}
  isRetryable: false,
  data: {
    error: {
      message: 'Model not found gpt-4o-mini-search-preview',
      type: 'invalid_request_error',
      param: 'model',
      code: null
    }
  },

AI SDK Version

5.0.98

Code of Conduct

• I agree to follow this project's Code of Conduct

[syeddhasnainn]

have you tried this snapshot?

gpt-4o-mini-search-preview-2025-03-11

[oboe-ori]

Yes, same error. You can reproduce it with the raw request too:

curl -X POST "https://api.openai.com/v1/responses" -H "Authorization: Bearer API_KEY" -H "Content-Type: application/json" -d '{"model":"gpt-4o-mini-search-preview","input":[{"role":"user","content":[{"type":"input_text","text":"Write a short poem about TypeScript"}]}]}'

This worked in an older version, before #5647

Is it possible this makes the request invalid somehow?

ai/packages/openai/src/chat/openai-chat-language-model.ts

Lines 250 to 263 in 502db42

	} else if (
	this.modelId.startsWith('gpt-4o-search-preview') ||
	this.modelId.startsWith('gpt-4o-mini-search-preview')
	) {
	if (baseArgs.temperature != null) {
	baseArgs.temperature = undefined;
	warnings.push({
	type: 'unsupported-setting',
	setting: 'temperature',
	details:
	'temperature is not supported for the search preview models and has been removed.',
	});
	}
	}

[oboe-ori]

I traced the older version, it works because it uses a different API endpoint:

v1/chat/completions is now v1/responses in the new versions.

curl -X POST "https://api.openai.com/v1/chat/completions" -H "Authorization: Bearer API_KEY" -H "Content-Type: application/json" -d '{"messages": [{"role": "user", "content": "Write a short poem about TypeScript"}], "model": "gpt-4o-mini-search-preview"}'