From December, ChatGPT and Codex share one allowance, measured in tokens.I checked where a week of my Claude Code usage went. 71% was the model re-reading cached context, 6% was output, and 57% happened in sessions already past 400k tokens. After I started fresh sessions sooner, that last number was 43% the next week.At what context size do you start over?https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits#LLM #Programming #SoftwareEngineering
From December, ChatGPT and Codex share one allowance, measured in tokens.I checked where a week of my Claude Code usage ...
