Another week, there is a new model of artificial intelligence ready for general use. This time, it is an anthropier with the introduction Claude 3.7 Sonata. The company describes its latest version as the first in the market "Hybrid thinking model," Meaning that the new version of Claude can answer a question almost immediately or take his time to work through step -by -step. As the user, you can determine the approach required by Claude, with a drop -down menu that allows you to determine "Thinking" You want to take.
"We have developed Claude 3.7 Sonnet with different philosophy of other thinking models on the market. Just as humans use a single brain for both fast responses and deep thinking, we believe that thinking should be an integrated ability of border models instead of a completely separate model," Man writes. "This unified approach also creates a smoother experience for users."
Anthropor is not explicitly named Openai, but the company is clearly captured by its competitor’s shot. Between GPT-4, O1, O1-MINI And now O3-MiniOpenai offers many different models, but unless the company closely follows, the number of systems offered can be overwhelming; In fact, Sam Tamman recently Admit the same amount. "We hate the selection of models as much as they do and want to return to Magic Unified Intelligence," he Posted on X Earlier this month.
Anthropor says it also took a different approach to developing Claude’s thinking capabilities. "We have improved somewhat less for competition problems in mathematics and computer science, and instead, we have focused on realistic tasks that better reflect how companies already use LLMS," The company writes. To this point, Claude’s current users can look at this "Especially strong improvements in coding and developing web in the front end."
Claude 3.7 Sonnet is available for use from all Clauds plans, including the free layer in humans. Meanwhile, developers can access the new form through the company’s applications interface, Amazon Bedrock and Google Cloud’s Vertex Ai.
When talking about developers, Antarbur also offers Claud Code, which is new "agent" The tool that allows you to delegate coding tasks to Claude directly from a terminal interface. Available as a limited research inspection, Antarbur says that Clood Code can read the code, edit files, write, run tests, and even pay to Gitap.
This article was originally appeared on Engadget on https://www.engadget.com/Ai/anthropics-new-clauds-model-can-think-both-fast-low-203307140.html?
[og_img]
Source link