OpenAI's API users get full access to the new o1 model
Briefly

OpenAI announced full WebRTC support for real-time voice APIs, simplifying audio interface development from 250 lines of code to around a dozen for third-party apps.
The new direct preference optimization method allows model makers to fine-tune AI by providing two responses and indicating a preference, enhancing the model's learning process.
Read at Ars Technica
[
|
]