← sources

Which is the way to go? Claude Code Max 20x or split the work between 5x and Codex Pro 100?

source: reddit (reddit:ClaudeAI)published 2026-10-29 20:18ingested 2026-10-30 02:36

retained text

I am looking for advice from people who are working on large projects and need large token limits.

I am working with a large project using Max 5x and purchased Codex Pro 100 to use simultaneously because it doesn't have 5 hour limits. However, I am finding that the code GPT Sol 6.1 writes is very ambiguous compared to Opus 5.5 (even though it works) and I spend a good amount of time getting Sol to the same standards I set for Opus.

To manage the differences I decided to spend some time to separate the work these models do and not get in each other's way, because Opus is good at what it does and is good at doing what it's told while Sol takes some liberties or 'forgets' some rules, is faster, but the code is not great (in my opinion) which in turn Opus takes as "this is committed code, therefore it is an approved way of writing, therefore I follow the same". Another factor is that OpenAI is still resetting the weekly limits, which for me can double or triple the 100 Pro token limits.

What is your recommendation, dear reader? What option did you go for?

document fetched 2026-10-30 00:21 · canonical https://reddit.com/r/ClaudeAI/comments/1xiuzcl/which_is_the_way_to_go_claude_code_max_20x_or