The 258K ceiling

OpenAI's GPT-5.5 gets a million-token context window through the API. Load the identical model into Codex, OpenAI's flagship coding agent, and live sessions report an effective window of 258,400 tokens against a published figure of 400,000. Developers reverse-engineered the number from Codex's own model catalog: it encodes GPT-5.5 at 272,000 tokens and shaves off 5 percent. 272,000 times 0.95 is 258,400, the exact figure sessions display.

The Codex model catalog caps GPT-5.5 at 272,000 tokens, so local overrides are clamped or ignored. — openai/codex issue #19464

The sharper detail is the direction of travel. The previous model, GPT-5.4, let Codex users push context up to a million tokens; upgrading to the model OpenAI markets as stronger at long context shrank the usable window inside the same tool by roughly four times. And the auto-compaction meant to rescue a long session by summarizing older turns is aimed at the advertised window, not the real one, so it fires too late and often fails outright, dropping the thread mid-task.

This is a narrow grievance with a wide lesson. The people who hit it are the heavy users: multi-file refactors, long debugging loops, large repositories, the sessions a coding agent is supposed to be best at. It surfaced not through a launch or a benchmark but through a cluster of GitHub issues where developers audited the config and did the arithmetic. The gap that matters is between the number a vendor advertises for a model and the number its own product silently enforces.