I've been using AI coding tools quite a lot lately, and there's one thing that keeps annoying me.
Usage limits.
I mean, these tools are genuinely impressive. You can build entire features, debug complicated issues, understand unfamiliar codebases, and do things that would've taken hours before.
But the moment you start working on something bigger, the experience changes.
You're halfway through implementing something, the agent finally understands your codebase, you've fixed a couple of issues, and then suddenly you hit a limit.
Now you either wait for it to reset, switch tools, or upgrade your plan.
And if you're someone who codes every day, this gets frustrating pretty quickly.
I've tried looking at different options, including Claude Code, Codex, Cursor, and a few others. They're all good at different things, but I kept coming back to the same question.
Why does using the more capable models for longer coding sessions have to be so expensive?
And why do I need to keep switching between different subscriptions just to use different models?
That's actually one of the reasons we started working on LucentraCode.
It's a CLI we're building that brings multiple coding models into one place, with a bigger focus on usage limits and affordability.
The idea is pretty straightforward.
If you're working on a backend, building an app, or spending hours debugging something, you should be able to focus on the work rather than constantly thinking about how much usage you have left.
We're including models like GPT-6 Astra, GPT-6 Sol, Claude, and others, with plans starting at $20/month.
We're also trying to offer as much usage as we sustainably can. That's been one of our biggest priorities from the beginning.
Of course, I'm not saying we've built something perfect. We're still improving a lot of things, and there are definitely areas where established tools are ahead.
But I think there's room for something that's more affordable and gives developers more freedom to experiment with different models.
If anyone's curious, this is what we're building:
Also, I'd love to know what other people are using right now.
Are you happy with the limits on Claude Code, Codex, or Cursor? Or do you also find yourself constantly watching your usage?
Genuinely curious because we're trying to understand how people actually use these tools beyond simple prompts.
Top comments (0)