OpenAI's Next Model May Be Fully Multimodal: Image and Audio Generation Under the GPT-5o or GPT-5.5 Name

Leaked details suggest OpenAI is building a successor to GPT-4o that natively generates images and audio — not just understands them — potentially arriving as GPT-5o or GPT-5.5.

OpenAI's next major model will reportedly be "the full garlic" — an omnimodal system that generates images and audio natively, similar to GPT-4o's multimodal input capabilities but extended to output, according to @mark_k. The model may ship under the name GPT-5o or GPT-5.5, positioning it as an evolution beyond the current GPT-5.2.

Unlock the full briefing

Get every story in today's briefing, the full archive, and the daily AI intelligence brief.

All stories today

Full archive

Daily brief

Cancel anytime. Payments powered by Stripe.