Fact Checked by 
Anthropic has this week (February 24) announced its newest AI and first hybrid reasoning model, Claude 3.7 Sonnet, which has an ‘extended thinking mode’ feature.
Described as being the “most intelligent model to date,” the new model can produce near-instant responses or extended, step-by-step thinking.
In comparison to its predecessors, the AI shows further improvements in coding and front-ed web development – with a command line tool for agentic coding, Claude Code, introduced too.
“Just as humans use a single brain for both quick responses and deep reflection, we believe reasoning should be an integrated capability of frontier models rather than a separate model entirely.
“This unified approach also creates a more seamless experience for users,” writes the Anthropic team in a news release.
Claude 3.7 Sonnet, Anthropic’s new AI model, has extended thinking
Claude 3.7 Sonnet is both a large language model (LLM) and a reasoning model in one, with users able to pick when they want the tool to answer normally or if they want it to think longer before answering.
“Some things come to us nearly instantly: “what day is it today?” Others take much more mental stamina, like solving a cryptic crossword or debugging a complex piece of code.
“We can choose to apply more or less cognitive effort depending on the task at hand.”
The choice of having the AI consider questions for longer is offered in an ‘extended thinking mode’ toggle which includes a ‘thinking budget’ to control how long Claude spends on a problem.
When in this mode, the model will expend more effort in coming to an answer and that process will be detailed for the user to see. In explaining why the visible thoughts have been brought in, Anthropic lists three benefits: Trust, Alignment, and Interest.
“Being able to observe the way Claude thinks makes it easier to understand and check its answers—and might help users get better outputs.”
Some of the team’s researchers “with math and physics backgrounds have noted how eerily similar Claude’s thought process is to their own way of reasoning through difficult problems: exploring many different angles and branches of reasoning, and double- and triple-checking answers.”
Featured Image: Via Anthropic News
