Anthropic has introduced a new artificial intelligence model, Claude 3.7 Sonnet, which allows users to choose between a quick answer to a simple question and a more complex analysis that mimics human thinking. This is the first hybrid AI reasoning model on the market, the company announced in its blog post.
The model became available to both free and paid users on Monday, February 24. However, free accounts will not be able to use the extra processing power to get answers to their queries for now.
Claude 3.7 Sonnet has also received a "chain of reasoning" feature - users can see the detailed process of answer generation, which will help to understand the model's logic and identify errors.
Anthropic is also actively developing AI agents. Claude 3.7 Sonnet integrates the "computer use" feature, which allows artificial intelligence to interact with the user interface, browse web pages, enter text, and press buttons. The company is also testing Claude Code, a new AI agent designed specifically for coding automation.
Recently, a number of companies, including OpenAI, DeepSeek, and Elon Musk's xAI, have developed systems that can spend more time computing before generating an answer. Anthropic, however, is betting on a balance between complexity of analysis and ease of use.
Experts predict that such approaches could become an industry trend. For example, OpenAI CEO Sam Altman previously announced that the company is working on a combination of GPT and o-series models that will automatically determine how much time to spend computing an answer.
