Tibo's Day 12 | Reuse subagent context for more effective usage
Shared context, longer caching and routing changes reduce repeated processing in Codex and Work. Benefits depend on what a task can reuse.

Usage resets, product updates, service incidents and practical guides for Codex and OpenAI users.
Shared context, longer caching and routing changes reduce repeated processing in Codex and Work. Benefits depend on what a task can reuse.
The update targets model-capacity blocks: fewer messages and improved continuation after congestion, without a fixed recovery-time promise.
Tibo first expected completion within two hours and later confirmed it. All three posts belong to October 14 in Pacific time.
ChatGPT and Codex desktop change how state data is stored, addressing heavy users’ disk usage. The 60% figure is the author’s expectation.
The posts are dated October 13 in UTC but remain Day 8 in Pacific time. The author explicitly describes this reset as global.
Codex and Work get a larger default context. Tibo also clarifies long-context charges when this model is used through a subscription.
Tibo first announced a global reset, then confirmed propagation. The messages do not announce an additional banked-reset balance.
The desktop Pro beta suggests a next message without charging usage for the prediction. A separate mobile update enables creating and messaging dots.