GPT-5.4 reportedly brings 1M-token context window and 'extreme' reasoning mode

OpenAI's next model, GPT-5.4, will feature a one-million-token context window (more than doubling GPT-5.2's 400,000 tokens) and an "extreme" thinking mode that burns significantly more compute on hard problems. The model should drop "sooner than you think," OpenAI teased on X, though no official date has been shared. The extreme reasoning mode targets researchers rather than everyday users, while the expanded context and improved reliability on multi-hour tasks are designed to benefit coding agents like Codex. The faster release cadence reflects OpenAI's effort to manage expectations after GPT-5's launch set an impossibly high bar.

View full digest for March 5, 2026