Tibo, Product Lead of OpenAI Codex: We've optimized Codex and Work to improve real-world performance. We achieved this by reusing context across sub-agents, extending context caching time, and improving request routing (thus increasing cache hit rate

2026-10-17

Tibo, Product Lead of OpenAI Codex: We've optimized Codex and Work to improve real-world performance. We achieved this by reusing context across sub-agents, extending context caching time, and improving request routing (thus increasing cache hit rate). Furthermore, we've significantly improved the performance of GPT-6 Pro for all Pro users.