Is anyone else dropping $$$ on vibe coding tools, or am I just burning cash?
I'm spending $200/month on Claude Code. Normal, or am I nuts? How much are you really spending? What tool is worth every penny, and what's way overrated?
Poll: How much are you spending per month on vibe coding tools? Options: $0–$50, $51–$100, $101–$200, $201+
75 views
Replies
I’m keeping it pretty lean right now, spending under $50/month.
Perplexity Pro (covered by my telecom provider, normally $20/month)
Lovable ($25/month)
What’s worked for me is using these tools in a smart way rather than just throwing prompts at them. For example, with Lovable, which has an absurd credit system(especally with the new agent mode), I make sure every prompt is refined first:
Draft the prompt in Perplexity → “create a prompt for Lovable to [do X].”
Run that through ChatGPT → “trim unnecessary details for Lovable.”
That way, I only feed Lovable concise, optimized prompts and can stretch the credits much further.
I mostly use Lovable for the first couple of iterations of a feature, then switch to Perplexity(Claude Model) for bug fixes and refinements. Works surprisingly well and keeps my costs low, at least in the early stages.
AI Clips
@anant_vardhan Love this strategic approach! Your prompt optimization workflow (Perplexity → ChatGPT → Lovable) is brilliant for maximizing those credits. The way you transition from Lovable for initial feature iterations to Perplexity for refinements shows real cost-consciousness. Question: How do you decide when a feature is "ready" to transition from Lovable to Perplexity for refinements? Is it based on complexity, functionality, or something else?
@yadavajay I usually move away from Lovable after the first couple of iterations, basically as soon as it implements the core idea I had in mind. After that, I rely on Perplexity for refinements, like:
Refactoring and modularizing the code. I’ll take the big file Lovable generates and feed it into Perplexity to get a cleaner, more maintainable structure.
UI/UX changes. It’s much easier to experiment with different iterations in Perplexity until I’m happy with the result.
Small feature upgrades or bug fixes.
So for me, Lovable is mainly for prototyping the initial feature, or when Perplexity just isn’t in the mood to follow orders. :)
I pay $20 / month for Claude Pro (Claude Code). Use Continue CLI with Qwen3 Coder (I have some Scaleway credits), Gemini CLI (free tier), Qwen Code (free tier). Very structured approach and works wonders for me. These are all vibe coded:
https://github.com/brainless/letsorder - this is live and usable as a demo micro SaaS vibe coded
https://github.com/brainless/brainless.in - live, my personal site
https://github.com/brainless/nocodo - marketing website live, product not live
AI Clips
@brainless This is impressive efficiency! Your multi-model approach with Continue CLI + free tiers is really smart cost management. I love seeing the actual results—that portfolio of live projects (especially letsorder as a working SaaS) proves the "vibe coding" concept beautifully. Your structured workflow clearly works! Question: With Continue CLI orchestrating multiple models, do you find certain models perform better for specific phases of development (like initial scaffolding vs. debugging vs. optimization)?
@yadavajay thanks! I keep updating this playbook: https://nocodo.com/playbook/
With Continue I use Qwen almost always, so I have not compared.
General boundary conditions and prompt helper - Chat GPT (Free)
Gritty code - Claude Pro ($20), can't be wasting the 5 hour restrictions getting the prompt correct!