OpenAI's Olivier Godement explained the release of new open-source audio and reasoning models as a strategic move to help enterprise clients keep sensitive data in-house while still leveraging the OpenAI ecosystem.