At last, GPT-5 has arrived and is offered at no cost to all ChatGPT users.
On Thursday, OpenAI revealed its most sophisticated and fastest model to date during a live-streaming event. In a media briefing, OpenAI CEO Sam Altman elaborated on the model’s features. “GPT-3 felt akin to conversing with a high school student… GPT-4 resembled a conversation with a college student, and GPT-5 marks the first instance of feeling like discussions with an expert on any subject,” Altman noted, comparing it to a “PhD-level authority.”
OpenAI asserts that GPT-5 exhibits improved proficiency in coding, mathematics, science, writing, and healthcare. Significantly, the company claims the new model has a reduced hallucination rate compared to GPT-3, which stood at 33 percent. The key point, Altman remarked, is enhancements that are both clear and nuanced. “I attempted to revert to GPT-4, and it was rather disappointing,” he added.
GPT-5 will become accessible today to ChatGPT users on free, Pro, and Plus plans, though it is not available at this moment. OpenAI indicates that free users will experience the strictest rate limitations, and ChatGPT will revert to a lighter version called GPT-5 mini once users reach their limits. Pro users will enjoy increased rate limits, while Plus users will benefit from unlimited usage and access to a more robust model named GPT-5 Pro. OpenAI has also introduced three model sizes for API users: GPT-5, GPT-5 mini, and GPT-5 nano.
Let the experimentation commence.
This is an evolving story, and we will amend this post as new details emerge…
GPT-5 consolidates OpenAI’s tools into a single interface
GPT-5 presents a refreshed user experience by combining various tools into one and widening access to OpenAI’s powerful AI resources.
Users who found the model selection process perplexing will no longer need to independently choose the appropriate model for specific tasks. GPT-5 resolves this issue by integrating reasoning abilities from earlier models into a cohesive experience. ChatGPT’s head, Nick Turley, highlighted that this is the first occasion free users will gain access to reasoning without needing to contemplate it, as “GPT-5 inherently knows when to analyze.”
“The overall feel of this model is very positive, and I believe that users will certainly perceive that,” Turley stated, “particularly everyday individuals who have not dedicated time to contemplating models.”
OpenAI illustrated a vibe-coding scenario in which GPT-5 developed a language-learning application featuring quizzes and games based on a brief natural language prompt. “[GPT-5] truly empowers everyone with the capability of creating beautiful and effective coding,” affirmed Yann Dubois, a researcher on GPT-5.
GPT-5 debuts following years of eager anticipation and reported challenges. It has been more than two years since OpenAI launched GPT-4, the model that enhanced ChatGPT’s abilities and spurred AI model adoption. Nonetheless, the rollout of GPT-5 encountered <a href="https://www.wsj.com/tech/ai/openai-gpt5-orion-delays-639e7693" target="_blank" data-ga-click="1" data-ga-label