La Derecha Diario Sam Altman, CEO of OpenAI, confirms that ChatGPT-5 will be 'like a real person.'
The CEO of OpenAI explained why GPT-5 will mark a new stage in the evolution of artificial intelligence
Sam Altman, CEO of OpenAI, previewed developments that mark a turning point in the history of artificial intelligence. Its next model, GPT-5, promises to be more accurate, more natural, and to have long-term memory.
The advancement impacts multiple aspects: voice and image integration, fewer errors, and customized versions. Everything suggests it will arrive before the end of 2025.
What's new in GPT-5?
The main differentiator of OpenAI's new model is that it will be able to remember what you told it in previous sessions. GPT-5 will have long-term memory, which allows it to maintain context as if you were talking to someone who knows you.
This eliminates one of the biggest limitations of current systems: the disconnection between each conversation. You will no longer have to repeat who you are or what you need.
Altman: "It's like a real person"
The statement is significant. Sam Altman referred to GPT-5 as the beginning of a new era. According to him, it will not only be more human, but also safer, more efficient, and much more useful in daily life.
The goal is no longer just "to impress." It is to create a reliable, practical system that adapts to what each user needs.
Total integration: text, voice, and image
GPT-5 will be able to combine different formats in the same conversation. Text, audio, and image will coexist naturally, leading to richer and more intuitive experiences.
This is key for complex tasks such as image analysis, audio transcription, or multimodal conversations.
Fewer hallucinations, more accuracy
One of the weaknesses of previous models was the so-called "hallucinations": incorrect answers delivered with confidence. GPT-5 reduces these errors thanks to better training and more advanced filters.
When will GPT-5 be released?
Although there is no official date, everything indicates that GPT-5 will arrive before the end of 2025. However, OpenAI doesn't rule out possible delays, as happened with previous versions.
More posts: