Market Rumor: GPT-5 May Launch as Early as Early August

OpenAI is reportedly gearing up to release its new flagship model GPT-5 in August, with a potential launch as early as early August if all goes smoothly, following CEO Sam Altman’s repeated hints of an imminent release.

Altman earlier this year noted GPT-5’s key change: integrating the GPT series with the reasoning-focused o-series, eliminating the need to pick task-specific AI models. Thursday’s leak also suggests mini and nano versions will debut alongside the full model.  

Full and mini versions of GPT-5’s integrated reasoning model will be available via ChatGPT and API, while the nano version will be API-only.

This unified model aligns with OpenAI’s strategy to build a “general artificial intelligence” (AGI) system. As a nonprofit, achieving AGI is a critical milestone, potentially forcing Microsoft to relinquish rights to OpenAI’s revenue and future models. However, AGI in the GPT-5 generation seems unlikely.  

Investors should note the release could shift due to development, server capacity, or competitor moves.

Earlier this month, OpenAI delayed an open-weight model release, citing “additional safety testing and high-risk area reviews.” Thursday’s leak claims this “open model,” similar to o3 mini with reasoning, will still launch by late July, marking OpenAI’s first open-weight model since GPT-2 in 2019, available on Azure and Hugging Face.  

Even without the leak, Altman’s recent hype—e.g., his July 19 post saying “we’ll soon release GPT-5,” calling it an experimental model fusing future tech—signals its nearness.

He ruled out a gold-medal IMO-level model for months ahead. In a Wednesday (July 24) podcast, Altman shared using GPT-5: “This morning, I tested the latest model, got an email with a tricky question, fed it to GPT-5, and it nailed it.”  

The AI’s power left the “ChatGPT father” baffled. Altman said, “I felt useless compared to AI on this. I should’ve managed, but couldn’t—AI did, and it’s a strange feeling.”