OpenAI has officially launched GPT-5, its most powerful coding model to date. Improvements on the previous models reportedly include an ability to answer questions much faster, with increased accuracy. In comparison to OpenAI’s previous models, GPT-5 is designed to behave more like an “active thought partner’, proactively flagging concerns and offering more nuanced answers. That is, a system where based on the user’s prompt, the model could either comply with or deny the request. While this can be cut and dry when it comes to an obviously malicious prompt, OpenAI noted that tech can struggle to manage actions when the message is unclear.