OpenAI on 7 August unveiled GPT-5, its most advanced generative-AI system, and began rolling it out immediately to all ChatGPT users—including those on the free tier. The model is also available to developers through the OpenAI API, with enterprise and education customers scheduled to gain access next week. Chief Executive Officer Sam Altman described GPT-5 as “the best model in the world” for coding and writing and “a significant step along the path to AGI,” saying the system delivers PhD-level expertise. OpenAI claims the model reduces factual errors by about 45 percent versus GPT-4o and now provides “safe completions” for sensitive prompts rather than blanket refusals. GPT-5 operates as a unified system that automatically routes queries between a standard engine and a deeper reasoning version. It ships in four variants—GPT-5, mini, nano and the extended-reasoning GPT-5 pro—while paid ChatGPT Plus and Pro subscribers receive higher or unlimited usage. Developers can choose among three API tiers priced from $0.05 to $10 per million tokens. The company says GPT-5 sets new internal records, scoring 94.6 percent on the AIME 2025 math test and 74.9 percent on the SWE-bench Verified coding benchmark. It also expands multimodal understanding, accelerates response times on Microsoft Azure supercomputers and improves performance on health-related queries. The release comes as ChatGPT approaches 700 million weekly users and intensifies competition with models from Google, Anthropic and Elon Musk’s xAI. Forbes reports OpenAI is exploring a secondary share sale that could value the San Francisco-based company at about $500 billion.