excited to announce the launch of gpt-live API. a few things to note: the novel architecture of the gpt-live model needed a new API, providing much more control to developers compared to previous speech-to-speech systems. we did specific training to further improve the controllability and instruction following of the model. it is offered at a very reasonable and competitive price of $0.05 per minute for the live model. generally, i think we are on a steep trajectory of improving cost, latency, and quality of realtime conversational models. this is our first API release toward that end with the new model series. we are investing actively and heavily to make this the best across our 1p products and provide maximum flexibility to developers via API. proud of the team for this release! please expect regular updates on this model series from now on!