. Our company is also using it internally, and the feedback has been quite good.
where they were mentioned.
I consulted Leo, an independent entrepreneur (Solopreneur), about his experience with Cursor. He said: "I've used it, but now bolt.new is better. Last time, I wanted Cursor to help me uniformly modify some code, which couldn't be done through search-and-replace. However, Cursor indicated that anything over 400 lines could not be processed. And I had 10 Python files, with the longest one being 10,000 lines."
However, today we will introduce Cursor first, and tomorrow we'll talk about bolt.new.
Functionality
Powerful auto-completion
Cursor features powerful auto-completion capabilities that can predict the user’s next editing actions. Once enabled, this feature runs continuously, providing multi-line code editing suggestions based on the user’s latest modifications.
Cursor can predict the user’s next intention based on recent modifications and generate corresponding code.
Cursor supports multi-line editing suggestions, allowing multiple changes to be recommended at once, significantly improving efficiency.
Even if the input is inaccurate, Cursor can automatically correct code errors.
Cursor can predict the next position of the cursor, helping users navigate their code more smoothly.
Cursor chat function
Cursor's chat function allows you to interact with an AI familiar with your codebase. This AI can view your current file and cursor location in real-time, so you can directly ask questions like “Is there an error here?” Use ⌘+Shift+L or "@" to add specific code blocks to the context, and press ⌘+Enter to converse with the entire codebase.
Use @Codebase or ⌘+Enter to ask questions about the codebase, and Cursor will search within the codebase to find relevant code snippets related to your question.
Use the @ symbol to reference code as part of the AI's context. Typing "@" allows you to browse all files and code symbols within a folder.
Click the image button below the chat box or drag an image into the input field to add visual content to the chat context.
Use @Web to get the latest information from the internet, and Cursor will search the web for you, providing up-to-date answers.
Click the play button at the top of the chat code block to apply the code suggestions directly to the codebase.
Use @LibraryName to reference popular library documentation, or use @Docs → Add new doc to add custom documentation.
Cursor shortcut ⌘ K functionality
⌘ K allows you to edit and write code with the help of AI. To edit code, simply select a code snippet, click "Edit," and describe the desired changes. If you want to generate new code, just press ⌘ K directly.
Use ⌘ K for quick editing or writing of code. After selecting a code snippet, press ⌘ K and describe the modification needs; if no content is selected, press ⌘ K directly to generate new code.
Use ⌘ K in the terminal to describe commands in simple natural language, and Cursor will convert them into actual terminal commands.
If you have a quick question about a part of the code, select the code snippet and click "quick question" to instantly get an answer.
Comparison
| Features | Cursor | GitHub Copilot |
|---|---|---|
| Code generation and completion | Provides comprehensive code generation, capable of creating entire functions or even multi-file structures, offering context-aware suggestions based on the project structure. | Focuses on line-by-line code completion and suggestions, relying more on immediate context and comments to provide code suggestions. |
| Natural language understanding | Excels at understanding and executing complex natural language commands, allowing developers to describe functionality in plain language and convert it into code. | Can generate code based on comments and context, but performs less effectively with complex instructions compared to Cursor. |
| Context awareness and multi-file editing | Offers stronger context awareness, understanding the entire project structure and relationships between files, featuring Composer-like functions that allow simultaneous editing of multiple files. | Primarily operates within a single file, with relatively weaker context awareness. |
| User experience and integration | Requires downloading a standalone editor, with varying learning curves depending on the user; familiar to VSCode users, but new features may take time to master. | Seamlessly integrates into existing IDEs like VSCode, with a relatively gentle learning curve. |
| Customization and flexibility | Provides flexibility in choosing AI models, including OpenAI, Anthropic, and custom models, allowing developers to tailor AI behavior to match project needs. | Uses fixed models developed by OpenAI, lacking the level of AI behavior customization offered by Cursor. |
| Pricing and accessibility | Offers a free version, a professional edition for $20 per month, and a business edition for $40 per month. | Personal plan costs $10 per month, free for students and open-source contributors. Business version is $19 per month, and enterprise version is $39 per month. |
| Model selection | Supports multiple models, including OpenAI's GPT series and Anthropic's Claude models, offering greater flexibility. | Initially only used OpenAI's GPT models, but recently announced support for models from Anthropic, Google, and OpenAI. |
| Model customization | Allows developers to choose the most suitable model based on project needs, providing a higher degree of customization. | Currently relies mainly on preset models, with relatively limited customization options. |
| Model updates | Regularly updates supported models to ensure developers have access to the latest AI technology. | Recently announced support for more models, showing a focus on supporting multiple models. |
| Multilingual support | Supports multiple programming languages, though it may not cover as wide a range as GitHub Copilot. | Supports a wide range of programming languages, leveraging GitHub's extensive codebase to provide multilingual code suggestions and completion features. |
| Community support and feedback | The community size is relatively small but growing gradually. | Has a large user base and active community, enabling developers to easily obtain support, share experiences, and receive feedback and suggestions from the community. |
| Performance and accuracy | Performs well in terms of code suggestion accuracy and contextual understanding, especially when handling complex projects. | Excels in handling common programming tasks, but may require more careful review and adjustment when dealing with complex languages or coding patterns. |
| Learning curve | May require some learning time due to the need to adapt to a new editor and its features. | Integrates into existing IDEs, making the learning curve relatively gentle, allowing developers to use it in a familiar environment. |
| Customized workflows | Provides a highly customizable coding environment, allowing users to adjust the interface and features according to personal needs, supporting multiple programming languages. | Based on OpenAI's GPT-4 model, it aims to provide developers with real-time code suggestions, with relatively low customization levels, and users cannot freely adjust its functions and interface as they can in Cursor. |
| Integration capability | Supports seamless integration with existing development tools and workflows, allowing users to access documents and other resources directly within the editor, reducing context-switching time. | Through tight integration with GitHub, it offers powerful version control and collaboration features, but due to its reliance on specific IDEs, it may face environmental limitations in some cases, affecting overall performance. |
| Code modification ability | Supports multi-line editing, inline editing, and intelligent predictions, making it easier to modify and optimize code effectively. | Primarily generates new code, with relatively weaker capabilities for modifying existing code. |