@effect/ai-openai add options to enable OpenAI specific built-in tools when using OpenAiLanguageModel

[lildesert]

What is the problem this feature would solve?

I would like to be able to specify more options when using OpenAiLanguageModel, like adding built-in tools. Here's an example function:

const findSomethingOnTheWeb = (input: string) =>
    Effect.gen(function* () {
      const languageModel = yield* AiLanguageModel.AiLanguageModel;
      return yield* languageModel
        .generateObject({
          prompt: input,
          system: String.stripMargin(
            `some instructions`,
          ),
          schema: Schema.Struct({
            test: Schema.Union(Schema.String, Schema.Null),
          }),
        })
        .pipe(
          OpenAiLanguageModel.withConfigOverride({
            temperature: 0.1,
            // tools ? 'web_search_preview'
          }),
        );
    });

What is the feature you are proposing to solve the problem?

Update @effect/ai-openai to allow more config options and be closer to OpenAI API.

What alternatives have you considered?

The current alternative I have is to use the lower level OpenAiClient effect API with code like

const response = yield* openai.client.createResponse({
      model: 'gpt-4.1',
      tools: [{ type: 'web_search_preview' }],
      input: `find something`,
    });