GPT-realtime model default reasoning

API
, ,
1

Hi Team

These are the realtime models we add for user selection (taken from your models list)

"realtime_models": \[
        {
            "id": "gpt-realtime-2.1",
            "name": "gpt-realtime-2.1",
            "reasoning": {
                "items": \[
                    "minimal",
                    "low",
                    "medium",
                    "high",
                    "xhigh"
                \],
                "default": "low"
            },
            "preview": true
        },
        {
            "id": "gpt-realtime-2",
            "name": "gpt-realtime-2",
            "reasoning": {
                "items": \[
                    "minimal",
                    "low",
                    "medium",
                    "high",
                    "xhigh"
                \],
                "default": "low"
            }
        },
        {
            "id": "gpt-realtime-1.5",
            "name": "gpt-realtime-1.5"
        },
        {
            "id": "gpt-realtime-2.1-mini",
            "name": "gpt-realtime-2.1-mini",
            "reasoning": {
                "items": \[
                    "minimal",
                    "low",
                    "medium",
                    "high",
                    "xhigh"
                \],
                "default": "low"
            },
            "preview": true
        },
        {
            "id": "gpt-realtime-mini",
            "name": "gpt-realtime-mini",
            "preview": true
        }
    \],

What is the default reasoning used by the model (assuming we dont specify a reasoning level?)

Thanks so much

God bless

2

Hi @nbo2,

For Realtime models that support reasoning, the default reasoning level is low when no value is specified. This is confirmed in the model announcement.

I also noticed that this default is not currently documented in the Realtime API reference. I’ll flag this with the team for review.

Thank you for bringing it to our attention!