OpenAI now reveals more of its o3-mini model’s thought process

In response to the pressure of opponents DIPSEC of the Chinese AI AgencyOpen is changing the way of its new AI model, O3-miniIts step -by -step communicates the “thought” process.

Thursday Declaration The company’s AI-powered chatbot platform ChattGPT will see an updated “thought chain” that has many more steps to the model “logic” step in the model and how it came to the answers to the questions. According to the Openai, the premium ChatGP plans that use O3-Mini in the “High Logic” configuration will also see this updated reedout.

“We’re introducing an update [chain of thought] It is designed to understand how the model thinks and is designed for 3-minute, “A spokesperson of the OpenAI tells Techcruch through email.” With this update you will be able to follow the logic of the model, keep you more precision and confidence in its reactions. “

Righty models like O3-Mini make them completely information-cheek before giving results, which helps them Avoid some problems That usually up the model trip. The trade off is the reasonable models take a little longer to come to the solutions — for a few minutes more than a few minutes.

DIPSEC R1 MODEA “logic” model, which is the process of the O3-Mini line, reveals the entire thought process, which is the logic of many AI researchers rational desired methods. In addition to making the model easy to study, the rational steps provide better user experience in certain circumstances, helping to indicate when the model can be right – or wrong – on the track.

Opeena decided not to show the complete rational steps of O3-Mini and its predecessors, O1 and O1-Mini for some reason. Instead, users only saw the summary of reasonable steps – SummaryThe

When we briefed the people before the release of O1-Preview, watching the bed live was usually “ah” moments that made it clear that it was a big deal. These are not raw beds but it’s a big step around and I’m glad we can share that experience with the world. https://t.co/72Zpprhmfk

– Noam Brown (@Polyinomial) Feb 6, 2025

Opina still does not show the entire rational steps of the 3-minute, but the agency says “it has found a balance”: and 3-mines can “think” freely “and then organize his” thoughts “in more detailed summations.

“For the improvement and protection improvement, we have added an additional post-processing step where the model reviews the raw discipline of thought, removes any insecure materials and then makes it easier to do any complex ideas,” said the OpenAI spokesperson. “Extra, this post-processing step enables non-English users to receive the discipline of thought in their mother tongue, creating a more accessible and friendly experience.”

A Reddit AMA last weekOpening Chief Product Officer Kevin Well indicated that the change was coming.

“We are working to show more than we are today – [showing the model thought process] It will be soon, “He said.” TBD is above all – all thoughts take the competitive patron, but we also want people (at least energy users), so we will also find the right way to balance it. “

There is an AI-centric newsletter on TechCrunch! Sign up here Every Wednesday gets it in your inboxThe