OpenAI unveiled new API updates at its Dev Day on Monday. The company introduced GPT-5 Pro, its latest language model, alongside the new video generation model Sora 2 and a smaller, cheaper voice model. These updates were part of a series of announcements designed to attract developers to the OpenAI ecosystem, including the launch of an agent-building tool and the new ability to build apps directly within ChatGPT.
The addition of GPT-5 Pro is expected to appeal to developers building applications for finance, legal, and healthcare industries. These sectors require high accuracy and depth of reasoning, according to OpenAI CEO Sam Altman. Altman also emphasized that voice capabilities are becoming essential as they are one of the primary ways people interact with AI. To support this, OpenAI is launching “gpt-realtime mini,” a smaller and cheaper voice model in the API that supports low-latency streaming interactions for audio and speech. This new model is seventy percent cheaper than the previous advanced voice model while promising the same voice quality and expressiveness.
Creators within the OpenAI developer ecosystem can now access Sora 2 in preview through the API. OpenAI released Sora 2, its latest audio and video generator, last week alongside the Sora app, a competitor to TikTok filled with short AI-generated videos. The Sora app allows users to generate videos of themselves, friends, or anything based on a prompt and share them via a TikTok-style algorithmic feed. Altman stated that developers now have access to the same model that powers Sora 2’s video outputs for use in their own apps.
Sora 2 builds on its previous generation by creating more realistic and physically consistent scenes with synchronized sound. It also offers greater creative control, from detailed camera direction to stylized visuals. For example, a user can take an iPhone view and prompt Sora to expand it into a sweeping, cinematic wide shot. One of the most exciting developments is how well the new model pairs sound with visuals, delivering not just speech but rich soundscapes, ambient audio, and synchronized effects that are grounded in the visual content.
Sora 2 is pitched as a tool for concept development. This ranges from creating a visual starting point for an advertisement based on a product’s general vibe to a Mattel designer turning a sketch into a toy concept. This example, provided by Altman at Dev Day, highlights OpenAI’s partnership with the Barbie-maker to integrate generative AI into the toy-making pipeline.