OpenAI expands o3-mini’s reasoning transparency for clearer insights

On Thursday, February 6, OpenAI announced that both free and paid ChatGPT users will now see an improved 'chain of thought' display, offering greater insight into how the model arrives at its answers. Premium subscribers using o3-mini in 'high reasoning' mode will also benefit from this enhanced transparency.

In response to pressure from rivals including Chinese AI firm DeepSeek, OpenAI is modifying how its latest AI model, o3-mini, presents its step-by-step reasoning process. 

Models like o3-mini conduct thorough self-checks before generating responses, reducing errors but slightly increasing response time, often by a few seconds to minutes. DeepSeek’s R1 model, a similar reasoning-based AI, fully discloses its thought process—an approach many researchers favor for its transparency and usability. This method allows users to assess whether the model is on the right track.

When we briefed people on 🍓 before o1-preview's release, seeing the CoT live was usually the "aha" moment for them that made it clear this was going to be a big deal. These aren't the raw CoTs but it's a big step closer and I'm glad we can share that experience with the world. https://t.co/72ZpPrHmfK

— Noam Brown (@polynoamial) February 6, 2025

Previously, OpenAI chose to limit the visibility of reasoning steps in o3-mini and earlier models like o1 and o1-mini, partly for competitive reasons, showing only summarised versions of the process.