GPT 5 update: Sam Altman unveils Auto, Fast and Thinking as response modes

In a recent update following the launch of GPT-5, Sam Altman, the CEO of OpenAI, announced exciting new features on August 13. Users can now select from three distinct response modes: Auto, Fast, and Thinking. The new 'Thinking' mode comes with a limit of 3,000 messages per week, with an option to switch to a more limited version, the GPT-5 Thinking mini, for those needing additional capacity. Altman expressed that the majority of users are likely to prefer the 'Auto' mode, yet he believes that the added flexibility will serve a specific segment of users well. One of the standout features of this update is the impressive context limit of 196,000 tokens, allowing the model to handle much longer documents and conversations effectively. He mentioned that these limits may be adjusted as user behavior evolves. In addition, GPT-4o has made a return to the model selection options for all paid users by default. Should OpenAI choose to phase it out in the future, Altman assured users that they would receive ample notice. Paid subscribers can now utilize a new toggle in the ChatGPT web settings to access additional models, including o3, 4.1, and GPT-5 Thinking mini. However, he clarified that GPT-4.5 remains exclusively available to Pro users due to its higher GPU resource requirements. Altman also hinted at upcoming changes to GPT-5’s personality, aiming for a warmer interaction style that avoids the more annoying traits seen in GPT-4o. He noted that one of the lessons learned from recent experiences is the importance of providing users with personalized model personalities. Last week, OpenAI highlighted that GPT-5 significantly enhances performance in coding and intricate problem-solving, asserting that it is advanced enough to be considered a PhD-level expert in its capabilities. The company proclaimed, “GPT-5 is our strongest coding model to date.”