Pro user feedback: need better free models for paying users

Ideas Feedback
1

Hi everyone,
I’m a Cursor Pro user, and I rely on Cursor daily for serious production work.

One thing that really stands out compared to other tools is how they continuously offer decent free models even to paying users (for example, a stable always‑free base model, or multiple free models you can fall back to for lighter tasks), while on Cursor this part still feels missing.

Right now, as far as I can see, the only truly free model inside Cursor is Grok Code Fast, which is nice as an experiment but often underperforms compared to the top models for real‑world coding on larger codebases. In many workflows it ends up being more of a backup than a reliable daily driver.

It would be great to have:

  • At least one or more stronger free base models that Pro users can use without worrying about burning credits for simple tasks.

  • A clearer “always‑free” baseline so that power users can offload trivial refactors, small utilities, and quick chat questions without dipping into their premium allowance.

  • Possibly a tiered model strategy: premium for high‑stakes work, and robust free options for everything else.

Other AI coding environments are already pushing hard on this combination of paid premium + generous free models, and it makes a big difference in perceived value for heavy users.

I really like Cursor and want to keep it as my main environment, so I’m sharing this as constructive feedback from someone already paying for Pro. Are there any plans to introduce better free model options for paying users, or to expand beyond Grok Code Fast as the only free choice?

Thanks!

2 Likes
2

Hey, thanks for the feedback.

The current Cursor pricing model is based on API costs of models. The Pro plan includes $20 usage, which is spent depending on the chosen model. After reaching the limit, you can buy more usage at the same API rates or upgrade to a higher plan.

Grok Code Fast is indeed the only fully free model. However, keep in mind:

  1. Slow pool: After the limit is reached, some models (GPT, Gemini, Claude, etc.) continue to work through the slow pool - with delays but no extra charges. Details are discussed here: What free models are actually available on Cursor?

  2. Composer-1: An available model for simple tasks, it uses fewer credits than premium models.

I understand your point about competitors. I’ll pass this as product feedback to the team. Which competitor models do you find most useful as an always-free baseline?