OPENAI unveiled new API updates to its development day on Monday, presenting GPT-5 Pro, its latest language model, its new model of generation of Sora 2 videos and a smaller and cheaper vocal model.
Updates were part of a series of advertisements focused on developers to court the OpenAi ecosystem, including the launch of an agent construction tool and the possibility of creating applications in Chatgpt.
The addition of GPT -5 PRO could use applicants for financing, legal and health care developers – industries that need “high precision and depth of reasoning”, by Sam Altman, CEO of Openai.
Altman also noted that vocal capacities will be essential in the future because it will quickly become one of the main ways that people use to interact with AI. To this end, Openai launches “GPT-Realtime Mini”, a smaller and cheaper vocal model in the API which supports low-latency streaming interactions for audio and speech. The new model is 70% cheaper than the previous advanced vocal model of Openai, but promises the “same quality and expressiveness of the same voice”.
Finally, the creators involved in the OpenAi developer’s ecosystem can now draw from Sora 2 on the preview of the API. OPENAI published Sora 2, its latest audio and video generator, last week alongside the Sora application, a Tiktok competitor filled with short videos generated by AI. The Sora application allows users to generate videos of themselves, friends or everything based on an invitation, and to share it via an algorithmic tiktok style flow.
“(Developers) now have access to the same model that feeds the superb video outputs of Sora 2 directly in your own application,” said Altman.
Sora 2 relies on its previous generation with more realistic and physically consistent scenes with synchronized sound and greater creative control – from the detailed direction of the camera to stylized visuals.
Techcrunch event
San Francisco
|
October 27-29, 2025
“For example, you can take the view of the iPhone and encourage Sora to extend it in a radical and cinematographic stroke,” said Altman. “But one of the most exciting things on which we worked is the way this new model sounds with visuals, not only speech, but rich sound landscapes, ambient audio, synchronized effects that are anchored in what you see.”
Sora 2 is presented as a tool for the development of the concept, whether it is a visual starting point for a announcement based on the general atmosphere of a product, to a Mattel designer transforming a sketch into a toy concept – an example of Altman supplied to Dev Day which highlights the agreement of Openai with the manufacturer of Barbie to bring a generative AI in the toy manufacturing pipeline.
