ChatGPT maker OpenAI launches its fastest and most innovative model GPT 5, CEO Sam Altman says: Users will feel like they're interacting with…

Microsoft-backed

artificial intelligence

(AI) startup OpenAI has launched GPT-5. The company claims that this new model is a significant step toward integrating AI into businesses. During the keynote of the launch event, OpenAI's CEO, Sam Altman highlighted the improved performance of GPT-5, describing it as smarter, faster, and less likely to make mistakes. Drawing a comparison with earlier versions, Altman said: “GPT-3 sort of felt like talking to a high school student,” while “GPT-4 felt like you're talking to a college student.” According to him, GPT-5 offers an experience closer to speaking with a PhD-level expert.In a blog post, the company describes GPT-5 as uniting and exceeding previous OpenAI developments, including those related to reasoning, agents, and advanced math capabilities. OpenAI states that GPT-5 offers advancements in accuracy, speed, reasoning, context recognition, and problem-solving. The company suggests that its full capabilities will be realised when businesses begin applying it to new use cases.

OpenAI’s GPT 5 model: Availability

The company has confirmed that ChatGPT Team customers can already start accessing GPT‑5, with the model launching for Enterprise and Edu customers next week. GPT‑5 is now available in the OpenAI API, and developers can also start using it. Team, Enterprise, and Edu customers will also soon get access to GPT‑5 Pro, a version with extended reasoning for even more reliable and detailed answers.

ChatGPT Plus subscribers ($20/month) can get higher GPT-5 usage limits than free users, while Pro users ($200/month) will receive unlimited access to GPT-5 and an enhanced GPT-5 Pro version with improved performance.

What’s new in OpenAI’s latest and most powerful model, GPT 5:

GPT-5 is OpenAI’s first “unified” model, designed to combine the reasoning strengths of the o-series with the quick response times of previous GPT models. Unlike GPT-4, which focused mainly on generating responses to user queries, GPT-5 expands ChatGPT's capabilities to include tasks like building basic software applications, managing calendar events, and drafting research summaries.The update also introduces a more streamlined experience. Instead of requiring users to adjust settings manually, GPT-5 uses a real-time routing system to determine whether to prioritise speed or more in-depth responses based on the context.As part of the rollout, ChatGPT is also getting some interface updates. Users can now choose from four new response styles—Cynic, Robot, Listener, and Nerd—which aim to adjust how the assistant replies without needing specific instructions each time.OpenAI claims GPT-5 delivers strong all-around performance, slightly outperforming rivals like Claude Opus 4.1, Gemini 2.5 Pro, and Grok 4 in coding and science-related tasks, though it underperforms in some benchmarks. It scores well on real-world coding ) and PhD-level science questions. It has improved accuracy in health-related queries, hallucinating far less than previous models (1.6% vs 12.9% in GPT-4o).GPT-5 shows reduced hallucination overall (4.8% vs over 20% in earlier models) and is better at identifying unsafe queries while avoiding unnecessary rejections. Safety research suggests it's also less deceptive, with improvements in honesty and transparency. However, performance is mixed on agentic benchmarks where it underperforms slightly on some tasks.