[Feature Request] support new SEARCH-capable OpenAI models gpt-4o-(mini-)search-preview

[DirkSchlossmacher]

🥰 Feature Description

On Mar,11 2025, OpenAI introduced 2 new models that provide web search as part of chat completions API.

Developers can now get fast, up-to-date answers with clear and relevant citations from the web. In the Responses API, web search is available as a tool when using gpt-4o and gpt-4o-mini, and can be paired with other tools or function calls.

We cannot simply add these models in NextChat with /app/constant.ts:

• because parameters frequency_penalty and presence_penalty are default parameters for OpenAI calls, but not supported yet for the below models.
• because interface RequestPayload has props beyond the common minimum of props across all OpenAI models
  see /app/client/platforms/openai.ts

const response = await openai.responses.create({
    model: "gpt-4o",
    tools: [ { type: "web_search_preview" } ],
    input: "What was a positive news story that happened today?",
});

console.log(response.output_text);

https://platform.openai.com/docs/models/gpt-4o-mini-search-preview
GPT-4o mini Search Preview
Fast, affordable small model for web search

https://platform.openai.com/docs/models/gpt-4o-search-preview
GPT-4o Search Preview
GPT model for web search in Chat Completions

Response format
Temperature
1.00
Max tokens
2048
Stop sequences
Enter sequence and press Tab
0/4
Top P
1.00
Frequency penalty
0.00
Presence penalty
0.00

🧐 Proposed Solution

Can we please reduce the export interface RequestPayload to only cover the common minimum of props across all OpenAI models, but not more?
see /app/client/platforms/openai.ts

... and dynamically add props if needed for some models?

📝 Additional Information

No response