Codex-Max
Trace OpenAI's GPT-5.1-Codex-Max agentic coding line: why it must keep running across context windows, what apply_patch and compaction solve, and how partner tools retune themselves to fit it.
GPT-5.1-Codex-Max is OpenAI's frontier agentic coding model, made for project-scale work rather than short chat turns. It spans several context windows by compacting state, and its prompting guide asks it to read surrounding code first, then land related edits in batches through apply_patch. Because it leans on shell habits, tools such as Cursor retune sandbox rules and reasoning traces around it.
