Heavy user's math: mixing $100 Claude, $200 Codex, $30 Grok beats a single max plan
codtv132 · reddit · 2026-09-30
A heavy user shares a hands-on accounting of frontier model subscriptions and how to combine them:
- Codex $200: limits burned through 2–3 days before the week ends
- Claude $20: Opus 5.5 ran 2–3 hours straight on large repos before hitting the 5-hour cap — great value, so upgraded to $100
- Fable on the $100 plan: genuinely productive and drains usage more gently than Astra; Opus 5.5 works well as its subagent, or alone when conserving quota
- Setup: $100 Claude + $200 Codex (dropping to $100) + $30 Grok + local 9B models, splitting workload across them
- Tip: keep master specs in Google Sheets and point project instructions back to them instead of relying on model memory
- Ecosystem lock-in (chats, image gen, projects) keeps him at the $100 Codex tier
Verdict: properly split across subscriptions, you get more useful work than one max-tier plan.
More from coding & agent
- Adding a Mac Mini to the Cloud So Coding Agents Can Debug iOS Keyboard Bugs — lucasmeijer · 2026-09-30
- Developer flags Claude Code's memory usage as seriously bad — Sauers_ · 2026-09-30
- AI agent spent 45 minutes failing at a signup the author did in 3 minutes — paulnovosad · 2026-09-30
- Warp Terminal adds ChatGPT sign-in, letting subscribers use Codex quota in agent CLI — vikvang1 · 2026-09-30
- Dev reports sol 6.1 still finding bugs after Opus 5.5, calling it a good sign — banteg · 2026-09-30
- pi coding agent takes the stage in live demo — mitsuhiko · 2026-09-30