So I finally decided to jump into the DeepSeek hype for this big freelance project I have starting next Monday and I am honestly super hyped to see how it handles heavy logic! Im torn between sticking with VS Code using the Continue extension or just switching over to Cursor entirely since everyone keeps saying its the best for ai stuff but idk if the subscription is worth it if I can just use my own API key in VS Code? I am on a pretty tight budget this month and I need something that wont lag my older macbook air while I am deep in it. Which setup actually feels more seamless with DeepSeek for someone who is kinda picky about speed?
Honestly, seeing people suggest Zed or just basic Continue is cool, but I gotta disagree politely if you are doing a heavy logic project on a budget. I have been super satisfied using the Roo Code Extension for VS Code lately. It handles the DeepSeek V3 API Tokens way better for actually building stuff rather than just chatting in a sidebar. Continue is fine, but Roo feels more like an actual agent that gets things done. Since you are worried about lag on that older MacBook Air, you should try these tweaks:
I was just thinking... which specific year is your Macbook Air? If its an older Intel model, you really need to watch the CPU usage. Ive been messing around with Zed Industries Zed Code Editor lately and it feels way snappier than VS Code for me.
Just catching up on this thread. I definitely agree that managing your RAM footprint is the most critical factor for an older Macbook Air.