OpenAI is preparing to release GPT-5, the first major overhaul of its flagship language model since 2023. Company chief executive Sam Altman is already testing the system internally, and developers briefed on the roadmap say the launch is imminent. The hybrid architecture is expected to alternate between rapid conversational replies and a slower, higher-precision reasoning mode, delivering stronger answers while cutting token usage and associated API costs. OpenAI is also adding native support for audio, images and browser-based autonomous agents, positioning GPT-5 as an all-purpose model for coding, scientific problem-solving and customer interaction. Early corporate adopters such as restaurant operator Papa John’s say each model upgrade has widened opportunities for workflow automation and personalised service, and analysts expect GPT-5 to intensify competition with rival foundation models from Google, Anthropic and Meta. The rollout is critical for OpenAI after the muted reception of GPT-4.5. The San Francisco-based company handles around 2.5 billion ChatGPT requests each day, counts roughly 700 million weekly users and is forecast to generate about $12 billion in revenue this year, yet it remains unprofitable as it pours capital into computing power and talent to pursue artificial general intelligence.