ChatGPT and Codex Usage Limits to Merge in December! A Summary of the Changes
機械翻訳 / Machine-translated
A
AI Friends
2日前・13分で読める
機械翻訳 / Machine-translated
@aifriends
AI Friends（https://aifriends.jp）のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
Have you been hitting the limit on ChatGPT or Codex more often lately? Starting in December, OpenAI plans to merge the two separate usage limits into one. This article breaks down what is changing, and what remains unclear.
Media outlet TokenPost reported on October 31, 2026:
For Plus, Pro, and Business users, ChatGPT and Codex usage will be counted together under a single token-based limit. The change takes effect in December.
Codex — OpenAI's AI tool that assists with programming tasks — can be accessed from ChatGPT, editors, and the terminal. Previously, users sometimes had to keep track of separate limits depending on how they used it. Those limits will now be unified.
OpenAI reportedly notified users of this change in advance via the purchase screen, though neither the start date nor the specific cap sizes by plan were disclosed at that time.
A token is the smallest unit used when AI processes text — roughly one to several characters in Japanese. Your usage is measured by how many tokens you consume.
Consumption varies depending on the model and the type of task. More powerful models and more complex tasks use more tokens.
For example, getting a short question answered uses very few tokens. On the other hand, loading a large program and asking the AI to revise it can consume a significant amount at once.
OpenAI has stated: "Instant chat will not draw from your usage."
This means light, everyday conversations can still be used freely as before. Usage from the shared limit is expected to come mainly from agentic features (where AI plans and executes tasks on its own) and heavier workloads like Codex.
You can check your combined remaining ChatGPT and Codex usage, along with the reset time, under Settings > "Usage."
Once you hit the limit, you can choose to wait for the reset, upgrade to a higher plan, or purchase additional credits for the relevant features.
According to TokenPost, the aim is to integrate the agentic features of "ChatGPT Work" into the main ChatGPT product.
The overseas newsletter AI Brief shares a similar view — though it should be noted this is the author's interpretation, not something OpenAI has officially confirmed.
We are in an era where a single AI handles chat, research, and code generation in one continuous flow. Keeping separate limits becomes confusing when those use cases overlap.
Codex is currently managed on a per-5-hour usage window. OpenAI's pricing page lists the following:
These figures are estimates, not fixed limits. Separate weekly limits may also apply.
How these limits will change after the December consolidation has not been announced — and that is the most important unknown right now.
The start date and per-plan cap sizes have not been announced.
OpenAI has said that "most users should not see more frequent resets than before," but without specific numbers to compare, there is no way to verify this.
On Hacker News, the following reactions were seen:
The same thread also included comments suggesting that a previously planned reduction to the Pro "20x" limit had been reversed. These are user comments, and we have not been able to verify the official wording.
Multiple commentary articles report that Anthropic's Claude already shares a single usage limit between chat and Claude Code, structured as a combination of a 5-hour window and a weekly limit.
Note that this information comes from commentary blogs rather than official documentation, so please check each company's official pages for precise details.
In that sense, OpenAI's move can be seen as aligning with broader industry trends. The key difference is that ChatGPT retains "Instant chat" — a space that does not draw from the shared limit.
No confirmed information is available regarding how this change will apply to users in Japan. A Japan-specific start date or pricing has not been announced.
To illustrate the potential impact, consider three scenarios:
The first is an office worker who uses ChatGPT to create work documents. If their usage consists mainly of light questions, Instant chat will likely cover most of it, and they may notice little difference.
The second is a freelancer who builds websites on the side. If they run large batches of Codex tasks at night to generate code, the tokens spent on chat-based document work earlier in the day will now come from the same pool. There is a risk of running out toward the end of the month.
The third is a small team of five using the Business plan. If one team member runs heavy tasks consistently, it becomes harder for the rest of the team to plan their usage. Regularly reviewing the Usage screen would be a good practice.
A. December has been indicated, but a specific date has not been announced.
A. Plus, Pro, and Business.
A. OpenAI has stated that Instant chat will not draw from your usage.
A. You can wait for the reset, upgrade your plan, or purchase additional credits for the relevant features.
A. Under Settings > "Usage," you can view your combined ChatGPT and Codex remaining usage and reset time.
For now, check your current Usage screen to get a sense of how much you are using. Once more details are released, be sure to confirm the cap sizes for your plan.
This article is a cross-post from AI Friends.
