# ChatGPT and Codex share one limit from December; Instant exempt

> Plus, Pro and Business seats will spend one allowance on chat and coding. On OpenAI's own figures, model choice decides how fast it runs out.

- Published: 2026-11-01
- Author: 28creation Desk
- Format: News
- URL: https://28creation.com/articles/chatgpt-codex-shared-usage-limit-december/

## The short read

**OpenAI estimates 5 to 45 Astra messages per five hours on Plus, against 350 to 3,000 on Luna.**

- OpenAI says ChatGPT and Codex will draw from one allowance on Plus, Pro and Business from December; Instant answers do not count.

### What you can do with this

> If your team uses Codex and ChatGPT on the same seats, default routine work to Terra or Luna and check Settings > Usage weekly, so chat time does not run dry.

---

ChatGPT and Codex will spend the same usage allowance, OpenAI told subscribers in an email that [BleepingComputer reported](https://www.bleepingcomputer.com/news/artificial-intelligence/openai-says-its-doubling-pro-usage-limits-and-bringing-work-features-to-chat/) on Sunday: "Your use of ChatGPT and Codex will draw from the same allowance." If people on your team write code in Codex and do research or documents in ChatGPT, a heavy coding week can now leave a seat short of chat.

[VGTimes](https://vgtimes.com/news/173095-openai-to-combine-chatgpt-and-codex-usage-limits-in-december.html) reports that OpenAI's [help article on usage limits](https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits) dates the merge to December and names the Plus, Pro and Business plans. That page refused our automated reader, so its wording here comes from those two reports.

## What changes, and what stays outside the pool

Quick answers in Instant mode are exempt. OpenAI's email calls it "Unlimited Instant" and says it does not count toward the shared limit, while more advanced models and complex tasks use more of the allowance, BleepingComputer reports. Combined remaining usage and reset times will show under Settings > Usage. Run out, and you wait for the reset, upgrade or buy credits.

Three things are still missing. VGTimes notes that OpenAI has named no exact switchover date, no size for each plan's shared allowance and no rate for converting chat and Codex use into one unit. Separately, OpenAI doubled the Work and Codex allowances on its Pro 100, Pro 200 and Pro 500 plans from October 29, per BleepingComputer.

## Why: chat is being folded into the agent meter

The merge extends a meter that already exists. OpenAI's [Codex pricing page](https://learn.chatgpt.com/docs/pricing), read today, says "ChatGPT Work usage inside ChatGPT uses the same pricing, credits, and usage limits as Codex." So the agentic Work mode already shares Codex's budget.

BleepingComputer adds that a combined Chat and Work experience goes to a small test group in the coming weeks, with Codex staying a separate experience in the ChatGPT desktop app. December brings ordinary chat under the same meter.

The pricing page has not caught up yet. It still says "Overall Work and Codex plan limits are unchanged" and "Pro plans currently have no five-hour limit."

## The model you pick sets the pace

The same page shows why the default model matters more once the pool is shared. On Plus and standard Business seats it estimates these local messages per five hours:

| Model | Messages per 5 hours | Credits per million output tokens |
|---|---|---|
| GPT-6 Astra | 5 to 45 | 1,250 |
| GPT-6.1 Sol | 15 to 160 | 250 |
| GPT-6.1 Terra | 35 to 285 | 125 |
| GPT-6 Luna | 350 to 3,000 | 12.5 |

Astra output costs 100 times as many credits as Luna output. OpenAI calls the message counts estimates and says cloud tasks use more than local messages, so treat the ranges as a guide. Plus costs $20 a month; Business costs $20 per user a month billed annually, or $25 billed monthly.

The trade-off, as we read it: a shared pool lets a light coder spend their unused Codex budget on chat. It also means nobody on a seat can tell from Codex alone how much chat they have left.

## What to do now

1. Set Terra or Luna as the default for routine Codex runs, and keep Astra for tasks that need it. On the table above, Astra gets at most 45 messages in five hours.
2. Move quick lookups to Instant, which stays outside the limit.
3. Check Settings > Usage once the combined view appears, and before long agent runs.
4. For developers who hit limits, the pricing page says local chats can run on an API key at standard API rates. That moves the cost to your API bill.
5. On Business, Edu or Enterprise with flexible pricing, the page says workspace credits can be bought; Enterprise and Edu plans on flexible pricing have no fixed rate limits at all.

## Sources

- BleepingComputer, "OpenAI says it's doubling Pro usage limits and bringing Work features to Chat", November 1, 2026: https://www.bleepingcomputer.com/news/artificial-intelligence/openai-says-its-doubling-pro-usage-limits-and-bringing-work-features-to-chat/
- VGTimes, "ChatGPT and Codex will share a usage limit starting in December": https://vgtimes.com/news/173095-openai-to-combine-chatgpt-and-codex-usage-limits-in-december.html
- OpenAI Help Center, "Understanding ChatGPT and Codex usage limits" (not readable by our tools): https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits
- OpenAI, Codex pricing (read November 1, 2026): https://learn.chatgpt.com/docs/pricing