Portuguese 
English 
Spanish 
3 min of reading 
0 comments 
OpenAI CEO Claims GPT-5 Will Arrive Between June and September, But Raises Doubts About Format and Even the Need for a New Version
The next major version of OpenAI’s artificial intelligence, GPT-5, is expected to be launched by September. This information was provided by the company’s CEO, Sam Altman, during an interview on the company’s new official podcast, The OpenAI Podcast. However, he made it clear that there is still no official release date.
Altman stated that the debut will likely happen “probably during the summer of the northern hemisphere,” which corresponds to the months of June, July, or September.
Still, he emphasized that the company is debating internally whether to continue with significant version leaps or to adopt a more gradual update model, as was recently done with GPT-4o.
-
Norwegian company 1X opened the first vertically integrated humanoid robot factory in the US in California, while China in Guangdong produces 10,000 units per year.
-
Billion-dollar industry created by artificial intelligence already allows video conversations with deceased relatives using digital avatars that recreate voice, face, and memory in a new frontier of technology.
-
Chinese artificial intelligence that independently solved in 80 hours an open mathematical problem posed 12 years ago by American Dan Anderson — and proved its own result
-
Artificial intelligence for robots with human-inspired hands advances and expands machine learning capabilities in the new generation of robotics.
During the conversation, host Andrew Mayne, who is also a former OpenAI employee, asked if users would notice a difference between GPT-4.5 and a potential GPT-5. Altman replied that this difference may not be clear.
He explained that defining what constitutes a new version has become more subjective. With technical advancements and the use of continuous updates, the concept of version has lost the rigidity it once had in the past.
Altman stated that, previously, the process was simpler: the company would train a model, release it, and then train another. Now, there are different possible paths, including the possibility of continuing to improve the current model without necessarily changing its name to GPT-5.
The OpenAI CEO also avoided confirming whether GPT-5 would be an entirely new model, a combination of techniques, or a system based on multiple models. The company has adopted a more cautious stance, avoiding raising exaggerated expectations about the new version.
Another important point discussed in the interview was the concept of AGI — Artificial General Intelligence. For Altman, the definition of AGI has changed over the past few years.
He said that if asked five years ago, he would have given a very different answer from the one he would provide today. According to him, many goals previously attributed to AGI have already been surpassed by current systems.
Nonetheless, Altman denied that AGI has already been achieved. He explained that the available models are still not capable of making scientific discoveries independently or driving significant advances in human research. For him, these are central criteria for defining superintelligence.
In a recent statement, OpenAI also asserted that the development of AGI will not be a single event, but rather a continuous process. This represents a shift from the company’s previous view, which treated AGI as a milestone to be reached at a specific time.
At the end of the interview, Altman spoke about OpenAI’s business model. He stated that, for now, there are no plans to include advertising in ChatGPT. According to the executive, this could undermine user trust. He said it would be a “moment of trust destruction” if the system’s responses were influenced by ads.
Even without advertising at the moment, Altman acknowledged that the company is exploring ways to generate revenue from free users. According to a leaked business plan, this could happen as early as next year.
With information from Olhar Digital.
Be the first to react!