HACKER Q&A
📣 WorldDev

What AI tool to use for coding in 2025?


I feel I am a bit late to the AI party. If I'm honest, I was reluctant to take the plunge, because I love coding so much.

But lately I have taken more and more to ask ChatGPT 5.1-Thinking for code, and I have to reluctantly admit that it spits out useful code. And yes, much much faster than me.

So I feel I now need to use a proper AI assistant to assist in coding.

At that point, I am mainly interested in using AI for small to medium flutter apps.

Can anyone with a bit of experience with these tools give me some recommendation as to what the best tools are?

As far as I can see there are 2 dimensions (Integration and AI Model), with the top contenders being:

1) Integration

    1a) VSCode fork (Cursor)

    1b) VSCode extension (Github copilot)

2) AI Model

    2a) ChatGPT 5.1

    2b) Claude Sonnet 4.5
My criteria are code success (meaning I want the setup to spit out code that actually solves the task) and privacy (probably a tall order, but I'd like a setup that does not openly send companies the inner details of my code, my secret keys, etc).

I also wonder about costs, in particular if I'll have to duplicate subscriptions (e.g. have a chatgpt subscription + a cursor one, or if I can use an existing chatgpt subscription with an integration solution without paying an additional subscription).

I would be grateful for any recommendation/feedback/random thoughts.


  👤 jpalomaki Accepted Answer ✓
Add OpenAI Codex extension to VScode. Setup the ExecPlan instructions as described in [1].

Then start by writing a spec.md file where you describe what should be built. Write like you would write to a smart developer.

Then use the highest thinking model available with prompt "Create ExecPlan for the task @spec.md and write it to file". It will think a while and create the file.

Take a quick look at the generated file. It may have some open questions or surprises you want to review and write some answers to.

For the implemenatation I usually switch to medium. Then request with something like "Implement @execplan.md". If it has numbered steps, it seems to help to say "Impelement steps 1,2,3,4,5 and 6 on @execplan.md" - this way the agent is more likely to complete the whole plan in one pass.

[1] https://cookbook.openai.com/articles/codex_exec_plans