OpenAI has released new models, including GPT-Realtime-2.1 and GPT-Realtime-2.1-mini, which are intended for low-latency voice agents in the API. However, there is some confusion about the specific models released, with one outlet reporting the launch of GPT-Live-1, a full-duplex voice model. The new releases from OpenAI are expected to provide lower latency and improved AI reasoning for developers.