OpenAI's Tibo posted a Day 12 update saying Codex and Work now give you more effective usage. He attributed the gain to three changes: reusing context across subagents, keeping context cached longer, and improved request routing that increases cache hits.
That is the full claim from the post. He did not spell out how large the increase is, or which plan tiers see it first. For how Codex usage is structured by ChatGPT plan, see OpenAI Keeps Codex's Weekly Caps Vague: What Each ChatGPT Plan Actually Gets.
