Hi. I am new to cursor. As Cursor uses third party Ai technology (GPTs or Claude), how is it different? And while working with cursor, can we switch between different Ai model without using API? While using free
Plan, can we still use our own API?
Hi @Asjad123 ,
I am writing this from mobile device, and not totally sure of the answers to all your questions, but this might get the conversation started with others who can answer in more detail.
Q1: They optimise the chat process for the ai-assisted coding process, see here for some light reading and insights into how they think
https://www.cursor.com/blog/instant-apply
On a simple level, if you have tried to code with Chat GPT etc, you will know there is lots of switching environments, re-uploading modified code etc. Cursor has lots of features that make the whole process more integrated such as inline code generation and refactoring (Ctrl + K), more in-depth chat (Ctrl + L) with several context options such as files, folders, websites or even your whole codebase, and multi-file creation and editing with Composer (Ctrl + I). You can also use Ctrl + K in the terminal to generate commands via natural language text.
Q2: Yes, you can choose different models:
https://docs.cursor.com/advanced/models
Q3: Iâm not sure if free users can use their own API keys, hopefully someone else can answer that question.
Here are some cost, usage and api related posts:
Some other info you might find of interest:
https://docs.cursor.com/miscellaneous/privacy
Are requests always routed through the Cursor backend?
Yes! Even if you use your API key, your requests will still go through our backend! Thatâs where we do our final prompt building
Thank you so much for your response.
Can you please tell what do you mean by âswitching environmentsâ?
And one more thing, do I still need to give very clear and detailed prompt in Cursor just as we have to give while using ChatGPT?
Hi @Asjad123 ,
By âswitching environmentsâ I meant that if you have previously been using Chat GPT or something similar to help you with coding, the process probably looked something like this:
- Start chat with Chat GPT in browser
- Paste, or upload, some code in the chat
- Ask some questions, get some responses
- Switch to your code editor and apply the changes
- Do some more programming for a while
- Switch back to the chat
- Upload your new code
- Ask more questions etcâŚ
So it would have involved âswitchingâ between browser and code editor and doing lots of manual work.
But with Cursor, the chat is in your code editor and it knows about all of your code base, even as you continue to make changes to it etc.
In regard to your second question, yes it still helps to think and communicate clearly about what you want to achieve, so that Cursor, and whatever LLM you choose to use, can give you the most relevant and efficient responses etc.
Here is a post with some more introductory information about Cursor: