@btibor91: A few updates and clarifications here - this is related to the GPT-5.2 Thinking model in ChatGPT (n...
A few updates and clarifications here - this is related to the GPT-5.2 Thinking model in ChatGPT (not the OpenAI API) - this is not related to the 40% faster GPT-5.2 in the OpenAI API (OpenAI has confirmed that updating Juice values is not how that speedup was achieved) - based on what I've heard from OpenAI, the updates to default reasoning effort for GPT-5.2 Thinking in ChatGPT were focused on latency and faster responses, while still offering the option to use higher reasoning effort to maximize intelligence for complex prompts (Extended on Plus/Business, and Heavy on Pro) - OpenAI has now updated the ChatGPT release notes confirming they lowered Standard and Light thinking time in January, but the Extended thinking setting for GPT-5.2 was unintentionally changed to be lower as part of that update and has now been fixed - the release notes also mention that they periodically adjust default thinking time based on ongoing experiments to find the best balance between answer quality and response speed, and that thinking time is not directly comparable across different models as each model is tuned independently