OpenAI Announces GPT-5 and Shares Details on GPT-4.5

OpenAI CEO Sam Altman has outlined the company’s plans for upcoming AI models, including GPT-4.5 and GPT-5.

In a post on X (formerly Twitter), Altman stated that OpenAI aims to simplify its product lineup and develop a unified AI system that seamlessly adapts to different tasks.

GPT-4.5 and the Shift Toward Smarter Models

The next step is the launch of GPT-4.5, internally codenamed Orion. According to Altman, this will be the “last non-reasoning model”, with training having already been completed in October 2024. However, development faced challenges due to high computational costs and data limitations.

Once GPT-4.5 is released, OpenAI will focus on merging its traditional GPT models with the "o" series, aiming to create a more adaptive AI that understands when deep analysis is needed and when a quick response suffices.

GPT-5: A Unified AI System

OpenAI plans to release GPT-5 as an integrated system, combining various company technologies, including o3, OpenAI’s most advanced model to date. As part of this transition, o3 will no longer be available as a standalone model. The company first introduced o3 in December 2024, followed by a lighter o3-mini version in January 2025.

Upon GPT-5’s release:

• Free ChatGPT users will have unlimited access to the new model, but with safeguards against misuse.
• ChatGPT Plus subscribers will experience GPT-5 at a higher intelligence level.
• Pro plan users will gain access to an even more powerful version of the model.

Altman did not specify exact launch dates for GPT-4.5 and GPT-5, only stating that it could take “a few more months”. OpenAI remains focused on unifying its models and improving their adaptability across various applications.