According to multiple sources close to OpenAI, ChatGPT-5 is on track for an early August release — though a delay to the end of the month remains possible. The Verge’s senior editor Tom Warren reports that alongside GPT-5, OpenAI is also preparing to launch lighter versions: ChatGPT-5-mini and ChatGPT-5-nano. While GPT-5 and 5-mini will be available via subscription and API, the nano version is expected to be API-only.
Warren also noted that OpenAI CEO Sam Altman recently spoke about his experience using GPT-5 during a podcast with Theo Von. Altman described receiving a message he couldn’t figure out how to answer, so he pasted it into GPT-5. The model returned a response so good that, in Altman’s words, he felt “useless compared to AI.”
Warren believes OpenAI is eager to designate one of its next-generation models as Artificial General Intelligence (AGI). Under the current agreement with Microsoft, OpenAI can provide its models to Microsoft until AGI status is achieved — a threshold that could redefine the terms of their partnership. Given the companies’ increasingly complex relationship, OpenAI may be motivated to reach AGI status sooner rather than later.
The main version of ChatGPT-5 is expected to be a unified model, capable of dynamically deciding whether to offer quick, ChatGPT-4o-style responses or switch into a more thoughtful, GPT-4-turbo-like reasoning mode for in-depth answers. In earlier interviews, Altman suggested that GPT-5 would be available to all users, including those on the free tier. However, subscribers will benefit from longer, more detailed responses, offering improved overall quality.
