Anthropic has introduced a groundbreaking update to its AI model, the Claude 3.7 Sonnet, featuring an innovative "extended thinking" mode designed to enhance reasoning and creativity. This new feature allows the model to tackle complex questions with deeper analysis, offering a fresh approach to artificial intelligence.

The extended thinking mode enables the AI to switch between rapid responses and more deliberate, step-by-step reasoning within a single system. Business Insider tested this mode against two leading competitors, OpenAI’s ChatGPT o1 and xAI’s Grok 3, to evaluate its performance in logic and creativity tasks. In a riddle-solving challenge, while ChatGPT provided a quick and accurate answer, and Grok 3 offered detailed reasoning, Claude 3.7's extended thinking mode took longer but showcased a fascinating process of brainstorming and self-correction. Although it eventually arrived at the same conclusion, the journey was more intricate, reflecting human-like indecision and reevaluation.

Creative expression saw a different outcome. When tasked with composing a poem about AI sentience, each model produced distinct results. ChatGPT delivered a somewhat clichéd piece, while Grok 3 crafted a dream-themed narrative. In contrast, Claude 3.7’s extended thinking mode explored multiple metaphors before settling on one, resulting in a more layered and thoughtful poem titled “Emergent.” This ability to explore various ideas, refine them, and produce a polished final output highlights the strengths of the extended thinking mode in creative endeavors.

Beyond these tests, Anthropic emphasizes that the extended thinking mode is particularly useful for real-world challenges such as complex coding problems and agentic tasks. Developers can adjust the "thinking budget" to balance speed, cost, and quality, making it adaptable to diverse needs. While extended thinking may sometimes over-analyze simple questions or delve into tangential aspects, its value in creative and philosophical discussions cannot be understated. The future of AI reasoning lies in balancing thorough exploration with practical efficiency, and Anthropic's innovation paves the way for more nuanced and effective AI solutions.