In response to pressure from its opponents, including the Chinese company AI Deepseek, Openai changes the way the younger AI model, O3-mini, communicates the “thinking” process of “thinking” step-by-step.
On Thursday, Openai announced That the free and paid users of Chatgpt, the AI Chatbot platform, will see an up -to -date “thought chain” that shows more than the “logic” steps of the model and how it came to answers to questions. Premium Chatgpt Plans subscribers that use O3-MINI in “High Reasoning” configuration will also see this updated reading, according to Openai.
‘We enter an up -to -date [chain of thought] For the O3-MINI designed to make it easier for people to understand how the model is thinking, “a spokesman for Openai told Techcruch by email.” With this update, you will be able to follow his reasoning model, giving you more clarity and confidence in his answers ”.
Models of reasoning such as O3-MINI control the events in detail before giving results, which helps them avoid some of the traps that normally travel to models. The compromise is that the models of reasoning take a little more time to reach solutions-usually seconds to minutes.
Deepseek’s R1 model, a “reasoning” model according to O3-Mini lines, reveals the full process of thinking, which many researchers argue that it is the preferred approach. In addition to facilitating the model study, reasoning steps provide a better user experience in certain situations, helping to indicate when the model can be on the right – or wrong – path.
Openai has chosen not to show the full steps of reasoning for O3-mini and its predecessors, O1 and O1-MINI, partly because of competitive reasons. Instead, users saw only summons of reasoning steps – summaries that were sometimes incorrect.
When we informed the people in 🍓 before the release of O1-Preview, watching the cot Live was usually the time “Aha” for those who made it clear that it would be a big deal. These are not the raw swings, but it’s a big step closer and I’m glad we can share this experience with the world.
– Noam Brown (@Polynoamial) 6 February 2025
Openai still does not show the full steps of O3-mini reasoning, but the company said it “found a balance”: O3-mini can “think freely” and then organize its “thoughts” in more detailed summaries .
“To improve clarity and security, we added an additional step after processing, where the model examines the raw chain of thinking, removing any unsafe content and then simplifying any complex ideas,” the Openai spokesman continued. “In addition, this step after editing allows non -English users to take the chain of thinking in their mother tongue, creating a more affordable and friendly experience.”
In a Reddit AMA last week, Kevin Weil, head of Openai’s products, hints that the change is coming.
‘We are working to show a bunch more than we show today – [showing the model thought process] It will be very, very soon, “he said. ”.
TechCrunch has an AI -focused newsletter! Sign up here to get it to your inbox every Wednesday.
