On Monday, Anthropic unveiled a sophisticated AI model designed to deliver quicker responses or exhibit its reasoning process step-by-step, aiming to secure a competitive advantage in the generative artificial intelligence sector. The launch of Anthropic’s hybrid model, which integrates various reasoning methodologies to tackle complex challenges more efficiently, occurs during a time of intense rivalry in AI innovation. U. S.

tech companies are competing fiercely against each other as well as Chinese firms like DeepSeek and Alibaba. The startup, supported by Amazon and Google, announced that the Claude 3. 7 Sonnet model is its most advanced iteration, set to be accessible across all Claude plans—Free, Pro, Team, and Enterprise. However, the “extended thinking mode” feature will only be offered on paid plans. When in extended thinking mode, the model engages in “self-reflection before responding, ” which enhances its abilities in math, physics, instruction adherence, coding, and a variety of other tasks, according to Anthropic. The San Francisco-based company noted that this hybrid reasoning model is intended to prioritize “real-world” applications over mathematical and computer science problems, reflecting actual business use cases for large language models. Additionally, Anthropic is introducing a limited-release preview of Claude Code, an autonomous coding tool designed to assist developers with various coding tasks. This tool allows developers to “delegate substantial engineering work directly from their terminal. ” An autonomous coding tool is an AI-driven software application capable of independently handling coding-related activities. While users can determine how much time and resources they wish to allocate for answering questions, the company's pricing structure will remain consistent with its previous models. Anthropic’s new model is more affordable than the competing OpenAI’s o1 model, priced at $3 per million input tokens and $15 per million output tokens, in contrast to OpenAI’s $15 and $60, respectively.