OpenAI Could Release GPT-5, the Next-Generation of Its Generative AI Model, Later This Year: Report

OpenAI could unveil the next generation of its artificial intelligence (AI) model, GPT-5 (Generative Pre-Trained Transformers 5), in the coming months, as per a new report. The AI model will arrive as the successor to GPT-4, which was officially released in March 2023 and currently powers its popular chatbot, ChatGPT. It is said that the fifth iteration of GPT will also bring new capabilities such as AI agents that can operate autonomously. Notably, OpenAI CTO Mira Murati also revealed recently that the company’s first text-to-video AI model Sora could arrive later this year, too.

According to a report by Business Insider, GPT-5 could be released around mid-year, which would be the months of June or July. The new AI model was reportedly shared with some of OpenAI’s enterprise clients who revealed that the new model is more capable and comes with significant enhancements over its predecessor. Unnamed by the publication, the sources called GPT-5 “really good” and “materially better”.

It is said that the new foundation model for ChatGPT will also arrive with new features. One such feature will be AI agents that can be asked to perform tasks autonomously, as per the report. AI agents appear to be similar to GPTs – trainable mini chatbots launched in November 2023 that that can be customised for specific tasks. However, unsupervised operability is a new skill which is reportedly being added. No further details on how these AI agents function is known.

The release timeline for the AI model is not set in stone, however. As per the report, OpenAI is still training the model, which means there is a possibility of delays as well. In an interview with Lex Fridman, CEO Sam Altman was asked about when GPT-5 might be unveiled and he replied, “I don’t know. That’s the honest answer”. But at the same time, he highlighted that the difference between GPT-4 and GPT-5 will be similar to the difference between GPT-3 and GPT-4.

Last week, Murati spoke about OpenAI’s another new project, Sora, that can create one-minute-long videos from simple text prompts. As per the company CTO, Sora can be publicly available in the coming months and be priced similar to DALL-E.