Anthropic, the artificial intelligence company founded by OpenNAi extrabajadores, presents Claude 3.7the first AI model that can produce both conventional results and a controllable amount of “reasoning” necessary to solve complex problems.
The company states that the new hybrid model will facilitate users and developers to solve problems that require a mixture of instinctive results and reflection. “The user will have a lot of control over behavior, for example, how long he thinks, and may exchange reasoning and intelligence with time and budget,” explains Michael Gerstenhaber, product manager of Anthropic’s AI.
The “hybrid” claude of Anthropic
Claude 3.7 also includes a “notes” that reveals the model reasoning process. The function, very similar to that of the Chinese ia Deepseek model, helps the user understand how a model works to modify or perfect the indications.
Dianne Penn, Chief of Research Product at Anthropic, reiterates that notes is even more useful when combined with the possibility of raising and lowering the “reasoning” of a model. For example, if the model has difficulties to crumble a problem correctly, the user can ask him to spend more time solving it.
Vanguardia’s companies focus more and more to ensure that the models “reason” before adversities as a way to increase their capabilities and expand their usefulness. OpenAI, the company that started the current boom From AI with Chatgpt, it was the first to offer its O1 reasoning model, in September 2024. Some time later, Sam Altman’s company introduced O3, a more powerful version, while Google offered its Gemini model, Flash Thinking. In both cases, users have to change model to access reasoning capabilities, a key difference with respect to Claude 3.7.
#Anthropic #launches #model #hybrid #reasoning #world
