Antarbur launches the first “hybrid thinking” model in the world

The difference between the traditional model and logic is similar to two types of thinking described by the Nobel Prize -winning economist Michael Kahinman in his book 2011 Thinking quickly and slowlyRapid and instinctive thinking -1 thinking and slower trading system -2 thinking.

The type of model that made Chatgpt is possible, known as a large language or LLM model, produces immediate responses to demand by inquiring about a large nervous network. These outputs can be remarkably smart and coherent but may fail to answer questions that require step -by -step thinking, including the simple account.

LLM can be forced to imitate deliberative thinking if instructions are directed to reach a plan that you should follow after that. This trick is not always reliable, and models are usually struggled to solve problems that require wide and accurate planning. Openai, Google, and now Anthropier is all used Method of learning known as reinforcement learning To get their latest models to learn to generate thinking that indicates the right answers. This requires collecting additional training data from humans to solve specific problems.

Penn says that the thinking of Claude receives additional data on business applications including writing code and unification, using computers, and answering complex legal questions. “The things we have made improvements are … technical topics or topics that require long thinking,” says Ben. “What we have from our customers is a great interest in spreading our models in the actual work burdens.”

Antarbur says Claude 3.7 is especially good in solving coding problems that require step-by-step thinking, which surpasses Openai’s O1 over some criteria such as Swe-Bench. The company today releases a new tool, called Claude Code, specially designed for this type of coding with the help of AI.

“The model is already good in coding,” says Ben. But “additional thinking will be good for cases that may require very complex planning – you look at the base of a very large symbol of a company.”