# Claude Code vs. ChatGPT Codex — Comparison Checklist

Companion to Chad Sheppard's video: https://youtu.be/8VEL6Lb1NjI
Guide: https://chadesheppard.com/claude-code-vs-chatgpt-codex

This is a new companion checklist, not the exact prompt used in the video.

## Before each test
- Save a clean copy of the same project for each tool.
- Write one clear task and identical acceptance criteria.
- Record plan, monthly price, model, and reasoning setting.
- Record the usage-window reset time and starting allowance.
- Note existing conversation length and attached files.
- Keep secrets and private material out of comparison notes.

## Test notes
| Measure | Claude Code | ChatGPT Codex |
| --- | --- | --- |
| Plan / price | | |
| Model / reasoning | | |
| Starting allowance used (%) | | |
| Ending allowance used (%) | | |
| Elapsed time | | |
| Follow-ups needed | | |
| Tests passed / result worked | | |
| Remaining problems | | |

## Interpret the result
- Compare the increase in usage, not just the ending percentage.
- Each percentage belongs to that provider's own allowance.
- Check whether both results actually meet your acceptance criteria.
- Repeat on different jobs before calling one tool a universal winner.
- API token prices do not establish subscription token budgets.
- Verify current prices, models, and limits with each provider.

## Chad's workflow from the video
1. Start everyday building with Claude Code.
2. Clear unrelated conversations or compact context you still need; preserve important notes first.
3. Use Codex for genuinely difficult problems when extra allowance is worth spending.

Chad reported 25% usage in Claude Code versus 100% in Codex in under ten minutes on one video project, with each plan costing him $20/month. Your results may differ.