8/13/2025, 9:17:12 AM
GPT-5, OpenAI's newest and most sophisticated large-scale AI model, was unveiled on Thursday.
Everyone, including those who are not paying for it, can now access GPT-5 thanks to the corporation. According to OpenAI, the model is “a lot more useful,” faster, and smarter, particularly in areas like writing, coding, and healthcare.
During a press briefing, OpenAI CEO Sam Altman stated, “I tried going back to GPT-4, and it was quite miserable.”
OpenAI, the pioneer of AI chatbot ChatGPT, is set to reach 700 million weekly active users this week and is in talks with investors about a potential $500 billion stock sale, following its rapid growth since its launch in 2022.
Moreover, OpenAI reported a lower hallucination rate and extensive safety evaluations for GPT-5, resulting in less frequent answer fabrication and 5,000 hours of testing.
Likewise, OpenAI plans to use “safe completions” in GPT-5, providing high-level responses within safety constraints that won't cause harm, instead of explicitly refusing to answer potentially risky user questions.
Besides, OpenAI's Michelle Pokrass, a post-training lead at the company, stated that GPT-5 has been trained to recognise task failures, avoid speculation, and provide clearer explanations of limitations.
OpenAI showcased how GPT-5 can be utilised for “vibe coding”, a method where users create AI-generated software using a straightforward written prompt.
Additionally, OpenAI has developed a web app for English speakers to learn French using GPT-5. The app features an engaging theme, flash cards, quizzes, and daily progress tracking. The AI-generated software has some rough edges, but users can customise it with background changes or tab additions.
OpenAI is launching GPT-5 to its Free, Plus, Pro, and Team users, marking the first time that Free users can access a reasoning model. Users with higher usage limits will have access to GPT-5 mini, while Pro users will have unlimited access. ChatGPT Edu and Enterprise users will receive GPT-5 approximately a week later.
Views on the Breakthrough of GPT-5
Microsoft is introducing GPT-5 to its products, including Microsoft 365 Copilot, Copilot for consumers, and Azure AI Foundry for developers to integrate AI models into third-party applications. Box, a file management company, has been testing GPT-5 on various data sets.
Notably, Box CEO Aaron Levie claims that GPT-5 is a “complete breakthrough” in AI models, as it can retain more information and use higher reasoning and logic capabilities to make decisions. Unlike previous models that struggle with complex maths or logic within long documents.
OpenAI is introducing three model versions, GPT-5, GPT-5-mini, and GPT-5-nano, each tailored to different cost and latency requirements for developers.
Furthermore, OpenAI has released two open-weight language models, GPT-5, for the first time since its 2019 release. These models are designed for easy customisation and runability by developers, researchers, and companies, while also aiming to appeal to a broader consumer audience.
Ultimately, Altman compared GPT-5 to having a team of Ph.D.-level experts available, stating that people are limited by ideas, but not really the ability to execute, in many new ways.
Read more news: