Snapshot of the GPT-4o Audio Preview model from December 17, 2024, for audio understanding and generation.