Cursor integrates Claude 3.7 Sonnet to greatly improve programming capabilities and user experience

Just now, Cursor AI announced that it has integrated Claude 3.7 Sonnet and updated the user interface to make it more concise and easy to use. In addition, Cursor has introduced a cross-chat conversation function, which can automatically summarize chat summary and inherit it into the newly opened chat window, further improving the user experience.

Earlier, Anthropic officially released its latest inference model - Claude 3.7Sonnet. With its innovative design philosophy and significantly improved programming capabilities, this model quickly became the focus of the industry. The core innovation of Claude3.7Sonnet is that it integrates quick answers and deep thinking capabilities to simulate the thinking mode of the human brain and provide users with a smoother interactive experience.

In practical applications, users can choose standard mode to quickly obtain answers, or switch to extended thinking mode, allowing the model to reflect on itself before answering. This model is particularly suitable for complex tasks such as mathematics, physics, and programming, and can provide more accurate and in-depth answers. Through the API, users can also accurately control the model's thinking budget, up to 128K tokens, thus finding the best balance between speed, cost and answer quality. Unlike traditional models, Claude3.7Sonnet focuses more on practical commercial applications rather than simply competitive performance.

In the field of programming, Claude3.7Sonnet has performed particularly well. In one programming test, Sonnet led the way with a high score of 70.3% for other well-known models, such as OpenAI's o1, o3-mini and DeepSeek R1, which scored only around 49%. This result shows that Anthropic intends to position Sonnet as a powerful coding AI, focusing on improving programming capabilities to meet developers' needs in complex code base processing and full-stack updates.

Claude3.7Sonnet is now fully online, supporting free, professional, team and enterprise versions, and available on Anthropic API, Amazon Bedrock and Google Cloud's Vertex AI. However, free users are currently unable to use the extended thinking mode. In terms of pricing, Sonnet3.7 is consistent with the previous model: $3 per million input tokens and $15 per million output tokens, including thinking tokens.

In addition to improving programming capabilities, Claude3.7Sonnet has also shown excellent performance in other fields. For example, in the TAU-bench test, Sonnet achieved an accuracy of 81.2% in the retail scenario and 58.4% in the aviation scenario, leading the other models in general. In addition, Sonnet has performed well in instruction understanding, reasoning ability, multimodal processing and code writing, especially after opening the extended thinking mode, its performance in mathematical and scientific problems has improved by leaps and bounds.