Cursor vs Claude Code – Looking for Community Feedback

Discussions
1

Hi everyone :waving_hand:
I’m currently using Cursor as my main tool, and I really like the IDE-level workflow.
However, I haven’t tried Claude Code yet, and I’m curious to hear from people who have used both.

Quick questions:

  1. In what cases do you prefer Claude Code over Cursor?

  2. Does Claude Code handle large codebases / architecture-level reasoning better in your experience?

  3. Do you use it as a primary tool or alongside Cursor?

  4. Is it worth investing time (and cost) to try Claude Code if Cursor already works well?

Would really appreciate any real-world experience :folded_hands:
Thanks in advance!

2

I am curious about Cursor Ultra vs Claude Code Max (200usd) plans when using exclusively Opus 4.5

I heard there was a temporary increase in claude code limits over the holidays, but I am more interested in regular limits.

Currently I use Cursor only because I prefer full IDE over extensions/terminal applications.

3

Like you I considered Claude Code, but today I decided to get the $200 plan instead of Claude Code’s $100-$200 one. I was on the $70 Cursor one before.

It was painful to do to be honest, but I like the experience of an IDE too and I think Composer-1 is a pretty good and quick model that works for me. I spend a lot of time just asking it to do something, and checking the diffs and queuing up new stuff while I wait for it to finish. I guess you can just run Claude Code as an editor window but to me its not the same.

I also like that you can queue up commands - I’m not sure if thats a thing on Claude Code maybe someone can correct me. Apparently its a limitation of a terminal which means you can’t do that. For me thats a big thing.

That said I am not really happy with a lot of things right now - for example the weird non git “red/green” thing although it has been less of an issue recently. Also the phantom lines when I close and reopen Cursor its been a thing for ages now. I really hope this gets fixed.

Also I do think the AGENTS.md thing is just a lot of hot air - I can’t definitely prove it but it never does anything. I wrote a thread about it and I simply don’t trust it.

3 Likes
4

I think a good option is Cursor Ultra + Codex via ChatGPT Plus for large monotonous tasks + Opus via Claude Code for cheaper use of Opus.

5

Model: Claude Opus 4.5

Performance and task complexity: Cursor lastest > CC lastest (Model Claude Opus)
Response speed and task processing speed: Cursor > CC
Price: CC cheaper than Cursor

I’m still using both in parallel, and I also use many other tools like Antigravity, Gemini CLI, and Codex, but Claude Code and Cursor are the best at the moment.

I suspect that Antigravity’s Opus model is a fake model, or that the Claude 3.5-4.5 model has been fine-tuned and publicly presented as Opus, because the quality of the answers is not as good as Opus thinking in Claude Code or Cursor (the cutoff time, from what I’ve seen, is somewhere around April 2024).

6

1. When to prefer Claude Code CLI? Use it for side projects or remote work on a VPS. It’s perfect for multitasking, you can work on two things at once without opening a second heavy IDE. Since I usually have a monorepo, Figma, and Teams open, using the CLI saves my machine from lagging.

2. Handling large codebases? This depends more on the model (like GPT-5.2) than the tool. I find Cursor handles most tasks well if you break them down. If a task is too big to break down, I switch Cursor to a high-reasoning model.

3. Primary or alongside? Cursor is my primary tool. I only use the CLI for terminal- remote environments. You can even run the Claude terminal inside Cursor if you want the best of both worlds.

4. Is it worth the investment? Yes. It’s worth the $40 total ($20 for Cursor Pro + $20 for Claude Code) to have better options. Always compare them so you know the pros and cons of each environment.

Pros & Cons:

  • Cursor: Best for Enterprise review and model variety (Flash 3 for speed, GPT-5.2 for logic), Tab completion, But, it’s heavy on RAM.

  • Claude Code CLI: Lightweight and fast for terminal tasks. However, there’s no UI for deep code reviews and option for models.

My Take: It’s about preference. I work at an Enterprise company, so I need an IDE to review every task. I’m not a Neovim user, so I don’t want to delegate everything to a terminal. I trust AI to ship code, but the IDE gives me the context I need to review it properly.

7

For me the biggest différence is how fast your code gets written, and how long you’re waiting doing nothing in front of your laptop.

I’m waiting less with cursor, and id say rhe quality is similar

8

Thanks for the discussion. The community feedback is really valuable.

A few notes on the issues mentioned:

Phantom lines after reopening: There’s an active thread about this issue here: Cursor creates phantom unsaved changes on every restart (requires discarding multiple files).

AGENTS.md not working: This is a known issue: AGENTS.MD doesnt seem to work reliably if at all. We recommend trying more specific rules, or using .cursor/rules instead of AGENTS.md.

On the comparison with Claude Code, thanks everyone for the detailed feedback on the pros and cons of each tool. It helps the team understand what to prioritize next.

9

The docs still refer to it - IMO it should be removed or given a warning in the cursor settings that it is not working.

Can you give me a screenshot of an example e.g. .cursor folder with configuration in cursor settings?

I really want to make sure this is working properly and the agent follows a simple set of rules in a file (or files).

I’m maybe exaggerating slightly but it literally doubles the work in large codebases removing unnecessary XML etc.

Thanks for the help @deanrie

10

Hello,
I use both tools in a 100% “vibe-coding” workflow, because I don’t know Python syntax. I do know how to code in C and C++ though, but today most APIs and libraries are in Python, so I have to use this language as well.

I mainly develop scientific applications, typically between 3 000 and 10 000 lines of code. My projects involve a lot of numerical computation: integrals, interpolation, and geometric calculations.

Since Cursor is older than Claude Code, most of the hype on YouTube and social networks seems to suggest that Claude Code is much better than Cursor. In my experience, this is completely wrong, and I’ll explain why. This is more based on practical usage than on formal benchmarks, so I may be wrong, but here is what I’ve observed.

1) Speed and model choice
Cursor is faster and gives easy access to very fast models for simple tasks, which is extremely useful for simple tasks.

2) Easy switching between models
Cursor makes it very easy to switch between small models for simple tasks and large models for difficult ones.
For example, for graphics-related problems, the only LLM that consistently solved my issues was GPT-5.2. Both Opus 4.5 and Sonnet 4.5 failed many times. For simpler tasks, I use lighter models. I don’t know whether Claude Code allows model switching, but by default it only provides Anthropic models.

3) Cost
Cursor also feels much cheaper to me. You can add DeepSeek, and you even have access to Grok, which is currently free.

My experience is that it is a great tools but we still needs a lot of time to control and correct what the AI does. I found dozens of mistakes in the code, even if I try to specify everything : conversion problems, integral bounds, confusion between times (UTC, local time…), matplotlib difficulties to put the legends at the good places, forget to do interpolation when needed and asked, forget to define general functions that are used many times even if asked…

There are some drawback in Cursor : I think that the models we use are not clearely defined. We don’t know what there are, if there are fine tuned or not, where they work…
The first which would give two options might win the game :
1/ possibility to run on personnal working station with small models. Even though personnal GPU might not have enough power, we have to learn to code with smaller models otherwise it would be too expensive. I would love to mount my own system, based on solar energy during the day for instance in order to reduce my carbon costs using AI. This is something really promissing.

2/ clarify the models and tokens we use for each task or questions. Especially smaller ones in order to reduce the costs and CO2 emission. We have to improve ourselves to know if using a small models for a task would work or not. We need metrics to estimate everything because we have along with LLM and agents to improve ourselves as well and don’t do everything in a black box

3/ options to limitate what can be down with application like Cursor and what we can do on our own. We use too many tokens to do something which is not really valuable with AI. For instance, if we break up the code in small parts and small files, that may reduce the tokens used.

1 Like
11

i personaly like to use mainly cursor and somtimes abit or claud code within cursor

12

Totally. The “red/geeen” thing is maddening.

1 Like
13

chatgpt5.2-codex-high in cursor is much better at reasoning and code writing than Claude code with opus. I use both, Claude code seems to excel at scanning large code bases looking for things and using agents and tools, but the code it generates is often bugging and incomplete. chatgpt5.2-codex-high is pretty flawless even with large context.