Open AI launches GPT-5, nano, mini and Pro — not AGI, but capable of generating ‘software-on-demand’

OpenAI has officially introduced its long-anticipated lineup of large language models (LLMs), featuring four distinct versions of GPT-5: GPT-5, GPT-5 Mini, GPT-5 Nano, and GPT-5 Pro. This release marks a significant upgrade from the previous GPT-4 model, which debuted nearly two and a half years ago. The new models are designed to cater to a variety of user needs, balancing speed, cost, and computational depth. GPT-5 will soon become the exclusive engine for ChatGPT, which serves 700 million users weekly. However, subscribers of ChatGPT Pro, who pay $200 monthly, can still opt for older models for a limited period. OpenAI has revamped its system to automatically determine the appropriate model based on query complexity. For instance, simpler inquiries will utilize the standard GPT-5 or Mini models, while more challenging questions will engage a special “GPT-5 thinking” mode that enhances reasoning, albeit at the cost of response time. In addition to the commercial models, OpenAI recently launched GPT-oss, a suite of free, open-source LLMs designed for offline use on personal devices like PCs and laptops. Notably, while GPT-5 represents a leap in capabilities, OpenAI has clarified that it does not equate to artificial general intelligence (AGI), which remains a long-term goal. Sam Altman, co-founder and CEO of OpenAI, emphasized that although GPT-5 showcases significant advancements in reasoning and generalization, it still lacks key elements of AGI, such as continuous learning once deployed. This clarification is important, especially in light of contractual agreements with Microsoft, which stipulate conditions under which OpenAI may start charging for access to its models. Benchmark tests reveal that GPT-5 is nearing human expert performance in various fields, including law and engineering. Altman compared the upgrade to transitioning from a pixelated display to a high-resolution screen, making it a compelling tool for users. Among its standout features, GPT-5 can generate functional web applications from a single prompt, showcasing its ability to produce complex software quickly and with minimal errors. OpenAI aims to democratize access to GPT-5, making it available across all tiers of ChatGPT, including free users, while higher-tier subscribers enjoy increased usage limits. The new system also introduces enhanced safety measures to reduce incorrect refusals and improve response quality. In tests, GPT-5 demonstrated a reduction in factual inaccuracies and showed significant improvements across academic and real-world benchmarks. Developers will benefit from upgraded API functionalities, including new tools for customizing responses and minimizing latency. Major organizations, such as Jet Brains and Notion, have already begun integrating GPT-5 into their products, citing improvements in productivity and output quality. OpenAI's latest models reflect a strategic shift towards practical applications and integration rather than flashy features. With GPT-5 seamlessly embedded into existing tools, it is poised to transform user experiences and developer workflows worldwide.