I tested Anthropic's Claude 3.7 Sonnet. Its 'extended thinking' mode outdoes ChatGPT and Grok, but it can overthink.

Anthropic has launched Claude 3.7 Sonnet, a hybrid reasoning AI model that can switch between quick responses and extended thinking. This new approach aims to integrate reasoning as a core capability rather than a separate feature. In tests by Business Insider, Claude's extended thinking mode showed strengths in creative tasks, producing more thoughtful and layered poems compared to competitors like OpenAI's ChatGPT and xAI's Grok. However, for logical reasoning tasks like riddles, the extended thinking mode was less effective, as it took longer to reach the correct answer without improving accuracy.

Overall, Claude 3.7 Sonnet's extended thinking mode is beneficial for creative and complex tasks, allowing for exploration and refinement of ideas. However, it may overanalyze simple questions and become less efficient in straightforward logical reasoning. Anthropic suggests that the mode is designed for real-world challenges, such as complex coding problems, where more extensive exploration can be valuable. The model has shown superior performance in software engineering benchmarks compared to some competitors.

Key takeaways

Anthropic's Claude 3.7 Sonnet introduces a "hybrid reasoning model" that can switch between quick responses and extended thinking.
In logic tests, Claude's extended thinking mode was slower and less effective compared to competitors like ChatGPT.
For creative tasks, Claude's extended thinking mode produced more thoughtful and polished results, outperforming competitors.
Claude 3.7 Sonnet scored higher than competitors in benchmarks like the SWE for real-world software engineering tasks.

I tested Anthropic's Claude 3.7 Sonnet. Its 'extended thinking' mode outdoes ChatGPT and Grok, but it can overthink.

Key takeaways

Discussion (0)