What to expect about the up-coming merge of ChatGPT and Codex usage
retained text
This is just my interpretation of the upcoming ChatGPT/Codex usage merge, but I think ChatGPT subscriptions could lose one of their biggest advantages where users had many "Thinking chat" to spare. **For Plus users** OpenAI previously announced up to 3,000 GPT-5 Thinking questions per week. This wasn't reaffirmed for newer models, but it shows how generous the allowance used to be. With token-based metering, longer prompts involving calculations and extensive analysis over many steps could become expensive. Suppose 5–10 such questions exhaust a five-hour allowance, representing around 15% of the weekly budget. That translates to roughly **33–67 questions per week**, potentially down from the ~~3000~~. **For Pro users, say those paying $100** We previously had a published allowance of 50 Pro questions per week. Now that Astra can handle complex tasks involving multiple agents, a single demanding question could potentially consume 5–10% of a shared weekly budget, leaving only **10–20 such questions per week** down from ~~50~~. We don't know the actual charging rates yet, but if this happens, much of the advantage of GPT-Pro disappears, especially when similar tasks can be handled through Codex. **Using up the ChatGPT allowance would also mean having no Codex allowance left to spare.** OpenAI says most users shouldn't experience more frequent resets, we will know in a month.
document fetched 2026-11-02 00:21 · canonical https://reddit.com/r/ChatGPT/comments/1xkud9n/what_to_expect_about_the_upcoming_merge_of