AI pioneers at OpenAI unveil their latest artificial intelligence innovation
In the rapidly evolving world of artificial intelligence, OpenAI has announced the launch of its latest model, GPT-5. This new software represents a significant leap forward in AI capability, promising to deliver improvements in intelligence, reliability, and versatility compared to its predecessor, GPT-4.
GPT-5 is designed to function as a "PhD-level expert," integrating the best capabilities of previous models into a unified system that automatically optimises responses based on task complexity. According to OpenAI CEO, Sam Altman, GPT-5 is like an "expert in every field with a PhD."
One of the key features of GPT-5 is its substantially reduced hallucinations. GPT-5 is about 45% less likely than GPT-4 to produce factual errors on typical queries, and up to 80% less likely when using its "thinking mode" with chain-of-thought reasoning. This makes it much more trustworthy for complex, open-ended, and high-stakes tasks such as healthcare, coding, and scientific problem-solving.
GPT-5 also boasts a unified model architecture with smart routing. Instead of manually choosing between different sub-models as in GPT-4, GPT-5 uses an automatic "real-time router" that selects the best internal model configuration for the question at hand, balancing speed and depth of reasoning seamlessly.
Enhanced reasoning and tool intelligence are another strength of GPT-5. It excels at chaining many tool calls reliably, handling tool errors, and carrying out long-running, complex agentic tasks at state-of-the-art levels, significantly outperforming previous models.
New verbosity control is another feature of GPT-5, allowing developers to specify the desired length and detail level of responses via a "verbosity" parameter (low, medium, high) for more tailored outputs. Users can also interact with GPT-5 using preset personalities like "cynic," "robot," "listener," or "nerd" to make conversations feel more natural and contextually appropriate.
GPT-5 is also faster and more efficient, delivering better performance with 50-80% fewer output tokens, enhancing speed and resource efficiency while maintaining or improving answer quality.
Microsoft is cooperating with OpenAI and plans to integrate GPT-5 into its products. The model was trained on Microsoft's cloud platform, Azure. In the future, ChatGPT will run on GPT-5 for free users.
The AI competition includes companies like Anthropic, Elon Musk's xAI, Google, Meta, and OpenAI is among them. The new software is called GPT-5, and it's expected to communicate at a higher level than its predecessor, GPT-4. As of the current release, ChatGPT has nearly 700 million weekly users.
Sam Altman stated that the "GPT-5 era" will be characterized by the ability to "create software on demand." The focus on reducing hallucinations and improving reasoning marks a step-change in trustworthiness, making GPT-5 much more suitable for professional and enterprise use cases. Overall, GPT-5 represents both a quantitative and qualitative advance over GPT-4, with OpenAI and partners emphasising its role as a more expert, flexible, and robust AI system.
Artificial-intelligence-powered GPT-5, designed to act as a "PhD-level expert," is set to revolutionize various fields with enhanced reasoning, tool intelligence, and faster performance. Its reduced hallucinations and improved trustworthiness make it more suitable for professional and enterprise tasks like healthcare, coding, and scientific problem-solving.
