OpenAI, GPT?5 coming in August: more powerful, integrated, and closer to human reasoning

After months of anticipation and some delays, OpenAI is preparing to officially launch GPT?5, the next generation of its language model, in early August. The announcement, reported by The Verge , was publicly confirmed by OpenAI CEO Sam Altman in an interview on Theo Von's This Past Weekend podcast. Altman also recounted a time when GPT?5 brilliantly solved a problem he himself didn't know how to tackle, describing it as a "perfect answer" and calling the moment "a personal breakthrough" that made him feel "useless compared to AI."

One step forward

GPT?5 promises to be a major step in the race for advanced artificial intelligence: it will integrate the logical capabilities of the o3 line, avoiding the current fragmentation between different models for different tasks (e.g., natural language on the one hand and problem solving on the other). It will be a "generalist" model but with specialized depth: capable not only of understanding and generating text, but also of reasoning, planning, and solving complex problems.

The new model will be available not only within ChatGPT (most likely initially for Plus and Team users), but also through the API, in the GPT?5, GPT?5 mini, and GPT?5 nano variants. These "light" versions will allow the model's power to be adapted to different use cases, from mobile devices to edge applications.

With GPT?5, OpenAI not only strengthens its leadership position in the field of generative AI, but also proposes a new paradigm: fewer models, but greater integration, more coherence, and more power. The combination of advanced language generation and reasoning capabilities opens up new possibilities in fields ranging from education to medicine, from programming to customer service.

Towards a more open AI

The launch of GPT-5 will also be an important testbed from the perspective of commercial adoption and competition with models from Google, Anthropic, and Meta. But, in parallel, OpenAI is ready to publish a new open-weight model by the end of July, the first open-weight model since GPT-2 (2019). According to sources cited by The Verge , it will be a model similar to o3 mini, with advanced reasoning capabilities, available on platforms such as Microsoft Azure and Hugging Face. While not fully open-source, this move aims to respond to growing pressure for greater transparency and interoperability in basic AI models.

The final goal

Altman's greatest ambition remains the development of AGI (Artificial General Intelligence), a system capable of learning and operating autonomously on any cognitive task. However, Altman has clarified that GPT?5, while a significant step forward, will not represent AGI. "It will still be many months after launch before this model reaches a higher level of capability," he stated. A formal declaration of AGI by OpenAI would have significant legal consequences: according to the agreements with Microsoft, the company led by Satya Nadella would lose the right to future revenues from OpenAI models once AGI is achieved.