Exposed Claude Sonnet 3.7 will expand the top and bottom windows

A few days ago, TestingCatalog reported that while analyzing the code for Anthropic's Claude Sonnet 3.7 model, it found that Anthropic may be expanding the context window for Claude Sonnet 3.7. Specifically, Claude Sonnet 3.7 will increase the context window from 200,000 tokens to 500,000 tokens. Specifically, the context window for Claude Sonnet 3.7 will be increased from 200,000 tokens to 500,000 tokens, and TestingCatalog notes that it is unclear whether this upgrade will be available only to enterprise customers. Launched in February, Claude Sonnet 3.7 is understood to be the world's first dual-mode hybrid reasoning model. Its Extended Thinking mode, which enables deep self-reflection, excels at complex tasks such as math, physics and programming. It is interesting to note that Claude 3.7 Sonnet's public reflection process is relatively objective and lacks personalization. Anthropic said that it has not trained the model's thinking process with standard roles, and hopes to give Claude the maximum freedom to think on its own. Anthropic believes that the so-called "thinking" process does not necessarily reflect the AI's internal decision-making logic, and therefore Anthropic will continue to publicize Claude's thought chain.