Audio transcribe with structured

[sbc640964]

Currently it doesn't allow me to use this:

$response = Prism::structured()
     ->using(Provider::OpenAI, 'gpt-4o-transcribe')
     ->withSchema($schema)
     ->withPrompt($inputTextPrompt, [$audio])
     ->withProviderOptions([
           'temperature' => 0.5,
           'language' => 'he',
     ])
    ->asStructured();

the Error reutrn

"The requested model 'gpt-4o-transcribe' is not supported with the Responses API."

Is there another option?

[sixlive]

We converted our OpenAI implementation to the Responses API. Currently there isn't support for Audio via the Responses API - docs.

What you could do is use Prism::audio to generate the transcription and then pass the transcription in the prompt for structured output.

[grpaiva]

@sbc640964 For now you can use openai-php/laravel to do that, it's pretty straightforward:

$response = OpenAI::audio()->transcribe([
    'model' => 'gpt-4o-mini-transcribe',
    'file' => $audio,
    'response_format' => 'text',
]);