[FEATURE] /fast mode should consume included quota instead of billing separately

Resolved 💬 4 comments Opened Mar 7, 2026 by nassimsyntree Closed Apr 10, 2026

Preflight Checklist

  • [x] I have searched existing requests and this feature hasn't been requested yet
  • [x] This is a single feature request (not multiple features)

Problem Statement

I'm on the Max plan ($200/month) and currently using only ~5% of my included quota. When I enable /fast mode, it bills extra usage on top of my subscription — even
though I have 95% of my normal quota unused.

This feels unfair: I'm already paying for a large quota that I'm not fully using, yet I can't use it to get faster responses.

Proposed Solution

/fast mode should consume the included plan quota at a higher rate (e.g. 2x tokens per message) instead of billing separately as extra usage.
Ideal behavior:

  • When I toggle /fast, it uses my existing Max plan quota — just faster and at a higher token cost per message
  • Extra billing only kicks in once my included quota is fully exhausted
  • This lets users trade quota capacity for speed within their existing plan

Alternative Solutions

Currently I just stay on normal mode and accept slower responses, even with 95% of my quota unused. There is no workaround — you either pay extra or go slow.

Priority : Important - Affects my productivity

Priority

Low - Nice to have

Feature Category

Performance and speed

Use Case Example

  1. I'm a Max plan subscriber ($200/month) using Claude Code daily for software development
  2. I've only used ~5% of my included quota this month
  3. I toggle /fast to speed up my workflow
  4. I get billed extra usage on top of my subscription — despite having 95% quota remaining
  5. I disable /fast and accept slower responses to avoid surprise charges
  6. With this feature, /fast would consume my existing quota at a higher rate instead of billing separately

Additional Context

Other services like GitHub Copilot and Cursor don't charge extra for faster responses within the same plan tier.

View original on GitHub ↗

This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗