Token usage increased significantly — same 00 budget does far less work

Summary

I'm a Max subscriber ($200/month). Since the recent changes (extra $200 usage allocation), my limits are being reached much faster than before with the original $200.

Before (original $200)

• Could create 200+ files daily
• Thousands of lines of code read and written per session
• Limit never reached even with heavy full-day usage
• Building a full multi-app SaaS system (8 apps, PostgreSQL, Node.js monorepo)

After (with extra $200 allocation)

• Limits reached within a few file edits
• A session that edits 5 files and runs some DB queries burns ~20% of daily allowance
• Simple fix sessions (read 3-4 files, edit 5 files, run build) consume disproportionate tokens
• Cannot sustain the same daily output that was previously possible

Example session (today)

• Task: Fix Instagram comment pipeline to stop unnecessary API calls (Apify)
• Work done: Read ~10 files, edited 5 files (~200 lines changed), ran TypeScript build, ran ~15 DB queries
• Result: 20% of session budget consumed for what should be a straightforward fix

Context

• Using Claude Code CLI on Windows
• Model: Opus 4.6 (1M context)
• Working on a monorepo with multiple CLAUDE.md files and memory system

Impact

This makes it very difficult to sustain daily development work. The value proposition of the $200 plan has decreased significantly — the same budget now covers far less actual coding work.

Ask

Please look into whether token accounting or model costs have changed in a way that reduces effective usage for paying subscribers. Transparency on what changed would be appreciated.