On August 7, OpenAI officially launched GPT-5, its next-generation flagship AI model that will provide ChatGPT users with even more powerful features and a cleaner user experience. GPT-5 is OpenAI's first "unified" AI model, combining the reasoning power of its o-series models with the fast response time of its GPT-series models, marking a new era for ChatGPT and its creator, OpenAI, and demonstrating OpenAI's broader ambition to transform AI systems from chatbots to intelligent agents.
GPT-5 not only provides smarter answers than GPT-4, but can also perform a variety of tasks on behalf of the user, such as generating software applications, managing schedules, or creating research reports. GPT-5 also simplifies the use of ChatGPT by eliminating the need for the user to manually select settings, as its built-in, real-time router automatically determines how to provide the best answer, whether it is a quick response or one that requires more "thinking" time. Its built-in real-time router automatically determines how to provide the best answer, whether it's a quick response or one that requires more time to "think" about.
In a briefing with reporters, OpenAI CEO Sam Altman claimed that the GPT-5 is "the best model in the world" and said that it is a "significant step" on the company's path to developing AI that can outperform humans in most economically valuable jobs (i.e., general artificial intelligence AGI). "important step." Starting Thursday, GPT-5 will be available to all free users as the default model, marking the first time OpenAI has made an AI inference model available to free users (Source.
The release of GPT-5 was highly anticipated, and since ChatGPT made OpenAI famous in 2022, ChatGPT has grown to become one of the world's most popular consumer products, with more than 700 million weekly users, close to 10% of the global population. Many industry observers see the GPT-5 as a bellwether of AI progress, and its acceptance in Silicon Valley could have far-reaching implications for large tech companies, Wall Street, and policymakers who regulate technology.
In terms of performance, OpenAI claims that GPT-5 achieves industry-leading levels in a number of areas, slightly ahead of leading AI models from Anthropic, Google DeepMind and xAI in key benchmarks. However, in some areas, GPT-5 performs slightly below other leading-edge AI models. For example, in SWE-bench Verified, a test of real-world coding tasks extracted from GitHub, GPT-5 scored 74.9% on its first attempt, slightly higher than Anthropic's Claude Opus 4.1 model (74.5%) and Google DeepMind's Gemini 2.5 Pro (59.6%).
GPT-5 also improved in answering health-related questions, with a much lower frequency of "hallucinations" (i.e., the tendency of AI models to fabricate information) than previous OpenAI models. In response to ChatGPT prompts, GPT-5 (with thinking) produced hallucinations and misinformation at a rate of 4.8%, significantly lower than previous o3 and GPT-4o models (22% and 20.6%, respectively).
The release of GPT-5 also brings a number of new features for both consumers and developers.ChatGPT users can now select four new personalities (Cynic, Robot, Listener, and Nerd) in their settings, which will automatically adjust the way ChatGPT responds without the user having to specifically ask the model to answer in a certain way. For developers, GPT-5 will be available in OpenAI's API in three different sizes (gpt-5, gpt-5-mini, and gpt-5-nano), and developers will also be able to control how verbose the AI model responses are.