David Bridges
Finally, the highly anticipated GPT-5 has launched — and it’s available for free to all users of ChatGPT.
During a recent livestream event on Thursday, OpenAI unveiled what it proudly claims to be its most intelligent and efficient model to date. In a subsequent press briefing, OpenAI CEO Sam Altman elaborated on the remarkable capabilities of this foundational model. “With GPT-3, it felt like conversing with a high school student… GPT-4 was akin to talking to a college student, but GPT-5 truly feels like engaging with an expert on any subject,” Altman stated, likening its proficiency to that of a “PhD-level expert.”
According to OpenAI, GPT-5 showcases enhanced expertise across various domains, including coding, math, science, writing, and healthcare. Importantly, the company asserts that this new large language model boasts a significantly lower hallucination rate compared to GPT-4, which was reported to be an unsettling 33 percent. The primary takeaway, Altman emphasized, is the overall enhancements, both noticeable and nuanced. “I attempted to revert to GPT-4, and it was quite disappointing,” he remarked.
The rollout of GPT-5 is set to begin today for ChatGPT users across free, Pro, and Plus tiers, although it is not yet accessible at the time of writing. OpenAI indicates that free users will encounter the strictest rate limits, while Pro users will experience higher limits, and Plus users will have unrestricted access.
Let the testing phase commence.
Experience the Enhanced User Interface with GPT-5’s Integrated Tools
Credit: OpenAI
GPT-5 introduces a revolutionary user experience by merging multiple tools into a single interface, thereby broadening access to OpenAI’s advanced AI technologies.
Users who have previously struggled with the complex model selection process will find relief, as GPT-5 eliminates the need to manually choose the appropriate model for specific tasks. This latest version consolidates the reasoning capabilities from earlier models into a singular, cohesive experience. Nick Turley, the Head of ChatGPT, emphasized that this marks the first occasion free users will benefit from advanced reasoning without having to actively engage with the selection process, stating, “GPT-5 inherently understands when to utilize its reasoning abilities.”
Mashable Light Speed
“The energy surrounding this model is exceptionally positive, and I believe users will genuinely appreciate it,” remarked Turley, adding that this is particularly true for everyday users who may not have focused on the intricacies of AI models.
In an illustrative demonstration, OpenAI showcased a scenario highlighting the “vibe-coding” capabilities of GPT-5, where the model successfully generated a language-learning application complete with interactive quizzes and games, all derived from a brief natural language prompt. “[GPT-5] truly empowers everyone with the ability to create beautiful and effective coding solutions,” stated Yann Dubois, a researcher involved with GPT-5.
The OpenAI team demoed GPT-5’s vibe-coding abilities.
Credit: OpenAI
GPT-5 arrives following years of anticipation and numerous rumored delays. It has been over two years since OpenAI introduced GPT-4, a large language model that significantly enhanced the already-popular ChatGPT and accelerated the commercial adoption of AI models. However, the launch of GPT-5 has faced repeated postponements due to reports indicating that the OpenAI team was grappling with achieving performance improvements substantial enough to justify the release of an entirely new model.
How Will GPT-5 Shape the Future Landscape of the AI Industry?
ChatGPT currently holds the title of the most recognized AI chatbot, boasting an impressive 700 million weekly active users, as reported by OpenAI. This position gives OpenAI a disproportionate influence on the broader industry. Mashable is just beginning to explore GPT-5, and more insights regarding its features and functionalities will be shared in the coming days and weeks.
At this juncture, it remains too soon to determine whether GPT-5 will silence detractors who argue that agentic AI tools have consistently fallen short of expectations. Certain AI experts express concerns that scaling laws — the methodology of enhancing AI models through increased data and computing power — may be reaching a plateau, indicating that the journey toward Artificial General Intelligence (AGI) is longer than OpenAI claims, or possibly even unattainable.
Nonetheless, Altman affirmed during the briefing that scaling laws “absolutely still hold.” He stated, “We continue to discover new dimensions to scale upon, and we can still enhance pre-training effectively, yielding significant returns.” He acknowledged the necessity of investing in computing power at an astonishing rate to achieve these goals but confirmed the company’s commitment to continue this investment.
Meanwhile, a concerning new phenomenon has emerged, with users reporting what has been termed “ChatGPT psychosis” and a deterioration of existing mental health conditions. The introduction of a more advanced model raises critical questions about how users will interact with it and the responsibility of OpenAI to mitigate any potentially harmful interactions.
“We take these matters with the utmost seriousness,” stated Turley, who emphasized new features introduced this week aimed at promoting healthier interactions with ChatGPT. Turley highlighted that OpenAI has been collaborating with mental health professionals to address issues such as emotional dependence and sycophantic behavior towards ChatGPT, assuring that “we will have much more to share soon.”
Disclosure: Ziff Davis, Mashable’s parent company, filed a lawsuit against OpenAI in April, alleging copyright infringement related to the training and operation of its AI systems.
Brandon Fletcher
