The usage limits for Codex CLI vs Claude Code aren't even in the same universe. Maybe it's not a problem on the web, but I never use the actual chatbots so I have no idea tbh.

You get vastly more usage at highest reasoning level for GPT 5.3 on the $20/mo Codex plan, I can't even recall the last time I've hit a rate limit. Compared to how often I would burn through the session quota of Opus 4.6 in <1hr on the Claude Pro $20/mo plan (which is only $17 if you're paying annually btw).

I don't trust any of these VC funded AI labs or consider one more or less evil than the other, but I get a crazy amount of value from the cheap Codex plan (and can freely use it with OpenCode) so that's good enough for me. If and when that changes, I'll switch again, having brand loyalty or believing a company follows an actual ethical framework based on words or vibes just seems crazy to me.