OpenAI has launched its “best AI system yet”, GPT-5. It goes so far as to say the GPT-5 is its “smartest, fastest, most useful model made”.
Along with the release of GPT-5, OpenAI also announced two new variants, GPT-5-mini and GPT-5-nano (in the API platform for developers). Generally, however, users will mainly access GPT-5 and GPT-5-mini.
The main difference between GPT-5 and GPT-5-mini is in its power, reasoning depth, and cost/efficiency. Where GPT-5 is like a large engine designed for demanding tasks, while GPT-5-mini is like a smaller engine built for routine, efficient work. However, some detail may be lost as it gives only a basic definition and skips deeper context to stay quick and clear.
What is GPT-5?
OpenAI calls GPT‑5 a “unified system”. What this means is that instead of the user having to pick between separate models for different tasks, GPT-5 automatically combines and switches between them behind the scenes.
For example, while it has a deeper reasoning model (GPT‑5 thinking) and a real‑time router, it will quickly decide on which to use based on what it knows about you, i.e. if you prefer fast responses or longer, more thought-out ones. GPT-5 brings extended reasoning for even more comprehensive and accurate answers.
Better in all ways
In testing, the company said that GPT‑5 not only outperformed previous models on benchmarks and answers questions more quickly, but significant advances in reducing hallucinations were made, along with improving instruction following, and minimising sycophancy, meaning that the model is less likely to simply agree with a user’s statements or opinions just to appear helpful or agreeable. It also performed better when writing, coding, and answering health-related questions.
In terms of writing, ChatGPT is now better at helping you with everyday tasks like drafting and editing reports, emails, memos, and more. In a demo, when tasked to create a poem, GPT‑5 was better at creating a more emotional arc with a stronger ending, clearer imagery, and more striking metaphors than GPT-4.
Talking back to you
OpenAI is also launching a research preview of four new preset personalities that can be concise and professional, thoughtful and supportive, or a bit sarcastic for all ChatGPT users. Initially available for text chat and coming later to Voice, the Cynic, Robot, Listener, and Nerd are opt-in, adjustable anytime in settings, and designed to match your communication style and how ChatGPT interacts with you.
Paid or free
Similar to previous releases, the difference between free and paid access to GPT‑5 is in usage volume.
While all ChatGPT users can access the new model, there is a cap on how much free users can use GPT-5. Once a free user hits that cap, ChatGPT will default to the less powerful GPT-5-mini.
For subscribers under the Personal plans, Pro subscribers get unlimited access to GPT‑5, and access to GPT‑5 Pro. Plus users can use it comfortably as their default model for everyday questions, with significantly higher usage than free users.
Business plan subscribers, Team and Enterprise customers can use GPT‑5 comfortably as their default model for everyday work, with limits that make it easy for the entire organisation to access it.