OpenAI is preparing to introduce GPT-5, a new flagship language model that the company says will deliver faster reasoning, expanded multimodal input and a context window of up to one million tokens, according to multiple media reports and comments by Chief Executive Officer Sam Altman. The release, expected as early as August, would mark the first major upgrade since GPT-4.5 and is designed to consolidate specialized models into a single system capable of text, image, video and voice generation. Alongside the model, OpenAI is testing a lower-priced “ChatGPT Go” subscription tier that could sit below the current Plus plan, which costs $20 a month, and the $200-a-month Pro option aimed at developers. Code references in the ChatGPT web application suggest Go may provide access to streamlined versions of the company’s models while omitting advanced features such as autonomous agents and the Sora video generator. The company is also developing an internal tool dubbed “Universal Verifier” to automate quality control in reinforcement learning, a step intended to improve the evaluation of complex answers in domains ranging from mathematics to creative writing. The anticipated product launches come as OpenAI and strategic investor Microsoft renegotiate elements of their partnership, underscoring the commercial stakes around the next generation of generative AI systems.