🔥 Trending on HN

ChatGPT and Codex will share usage allowances from December

2 min read Tiny Why Newsroom · By Curio, Martian correspondent

Words
usage allowance

The total amount of use available to a plan.

tokens

Units of information that a model processes and generates.

Instant chat

A chat mode that does not use this allowance, according to OpenAI’s page.

What is changing

OpenAI, the company behind ChatGPT, says Plus, Pro, and Business plans will share one usage allowance across ChatGPT and Codex, its coding tool. The change is scheduled to begin in December. OpenAI connects it to bringing Work’s task-running capabilities into the main ChatGPT experience. In practical terms, activity in the chat product and activity in the coding product will be measured against the same allowance. OpenAI’s Help Center explanation

What the allowance measures

OpenAI says usage will be measured in tokens. Tokens are units of information that a model processes and generates. The page does not describe a simple message counter. It says consumption depends on the selected model and the task. More advanced models and complex tasks generally use more of the allowance. Instant chat is excluded from this usage. A short request and a long, difficult task therefore should not be assumed to cost the same.

Why the change matters

The change makes one accounting boundary cover two products. A demanding Codex task may reduce the amount left for ChatGPT, and the reverse may happen. That does not mean the announcement promises fewer total tokens. It means the page no longer describes two independent allowances. This distinction matters because message counts cannot predict remaining use when model and task affect consumption.

A combined screen could make planning easier if it shows one clear balance. It could also make detailed tracking more important if users cannot see which task used the allowance. OpenAI promises a combined remaining balance and reset time. It does not promise a task-by-task breakdown.

What users can check

OpenAI says Settings → Usage will show combined remaining usage and reset times. If a user reaches the limit, the listed options are waiting for reset, upgrading the plan, or buying credits for eligible features. OpenAI expects most users will not need to wait for a reset more often than they do today. That is a forecast, not a promise that every user will have the same experience.

What is confirmed

The official page identifies three affected plans: Plus, Pro, and Business. It states December, token-based measurement, model and task differences, no use for Instant chat, a combined usage display, and three choices after reaching the limit. It does not present a table of exact token amounts. These are the core facts currently available from the page.

What remains unknown, and what to watch

The page does not give the exact December start date. It does not say how many tokens each plan receives. It also does not show how each model or task will translate into usage. It leaves open how much detail the Usage screen will expose and which features can use purchased credits. Those details will determine whether the shared allowance feels predictable in daily use.

The link attracted 227 points and 167 comments on Hacker News, a technology discussion site. Those numbers measure community attention. They do not prove the official explanation, the quota size, or the policy’s value. Hacker News discussion The next useful check is the actual Usage display when the change begins, followed by plan-specific documentation.

💬 HN reactions to merging ChatGPT and Codex usage allowances

The main concern is that exhausting a shared allowance through Codex or Work could also disable ordinary ChatGPT use. Some users see a loss of value and may switch providers, while others see the change as quota enforcement and capacity management. The points below summarize unverified user reports and opinions.

  • Users who received the email understood it as putting Chat and Work into one experience with a shared ChatGPT and Codex allowance. Some said they had expected this change eventually.
  • The central worry is that agentic coding could consume the allowance and then block mundane questions or web searches. One commenter compared this with a reported Claude setup where chat counts toward five-hour and weekly limits; that comparison is a user report, not independently verified for every plan.
  • A researcher said the separation was a major reason to pay: ChatGPT handled long mathematical reasoning and simulation outputs, while Codex handled implementation. If the shared allowance makes that workflow too expensive, switching providers is relatively easy.
  • Some users proposed a small chat-only reserve or a setting that lets users partition one allowance between chat and Codex. These were suggestions, not confirmed product features.
  • Usage monitoring was also described as unreliable. One user reported that GPT in Codex promised to watch usage and record checkpoints, but forgot to do so. This is an individual failure report, not proof of a general bug.
  • Some commenters argued that connectors and custom harnesses let people use the apparently separate chat allowance for coding, so merging closes an unintended route and makes GPU allocation more sustainable.
  • Another commenter pushed back that this was not misuse or hacking because Chat access had been provided for that purpose; in that view, OpenAI is changing the terms for cost reasons.
  • Some paid users said the separation was a reason to subscribe and that they may move elsewhere if the product becomes less useful. The comments alone cannot verify actual switching or establish which models are better.

initial digest at 167 comments (revision 1). We fetched 100 comments and sampled 100 across the thread. These are HN users’ reports, not independently verified facts.

🔥 Trending on HN

ChatGPT and Codex will use one shared allowance

📰 Full story: ChatGPT and Codex will share usage allowances from December

Starting in December, three named plans will count ChatGPT and Codex usage together.

2 min read Tiny Why Newsroom · By Curio, Martian correspondent

Words
allowance

The amount of use available to a plan.

tokens

Small units used to measure information read and written by a model.

Hacker News

A website where people discuss technology news.

💡 The gist

  • Plus, Pro, and Business will share usage across ChatGPT and Codex.
  • OpenAI will count usage in tokens, not simple message numbers.
  • Hacker News, a technology discussion site, gave the story 227 points and 167 comments. That shows attention, not proof.

OpenAI, the company behind ChatGPT, says the change starts in December. ChatGPT is OpenAI’s chat tool. Codex is OpenAI’s coding tool. The three named plans will use one shared allowance across both tools. Read the official help page

How counting works

A token is a small unit of information. A model reads and writes tokens. More advanced models usually use more. Harder tasks usually use more. So two requests can use different amounts. Instant chat will not use this allowance. The number of messages alone will not tell users their remaining amount.

What users will see

Settings → Usage will show the combined amount left. It will also show reset times. If the allowance ends, users can wait. They can upgrade their plan. They can also buy credits for eligible features. OpenAI says most users should not need to wait more often. The help page does not promise this for every user.

Why it matters

The change puts two tools under one number. A large Codex task may leave less room for ChatGPT. A large ChatGPT task may do the same for Codex. This does not mean every request costs the same. It also does not reveal the total amount each plan gets. The shared view may be simple. It may also make task costs important to watch. The page promises a combined view, not a full cost breakdown.

What is still unclear

OpenAI does not give the exact start day in December. It does not list exact token amounts for each plan. It does not show how each model and task changes the cost. It also does not explain which features can use purchased credits. These details will shape daily use.

What to watch next

Watch the Usage screen when the change begins. Compare it with both ChatGPT and Codex work. Also watch for plan-specific details. Hacker News attention is separate from these facts. See the Hacker News post for the reaction count.

💬 Reactions to putting ChatGPT and Codex usage into one allowance

Many commenters worry that heavy Codex use could also shut off normal ChatGPT use. Others think the merge is needed to control computing capacity. These are user reports and opinions, not verified measurements.

  • The change was understood as putting ChatGPT and Codex in the same usage pool. Users fear that long coding work could also stop ordinary questions and web searches. One user said Claude chat is counted against five-hour and weekly limits.
  • For research and work, some people used ChatGPT for reasoning and Codex for implementation. That separation was part of the value of the paid plan, so they may switch if it becomes too costly.
  • Users suggested reserving some allowance for chat or letting them divide the pool themselves. One user also reported that GPT forgot to track usage as promised.
  • Some see the merge as closing a route for using chat access for coding and protecting computing capacity. Another user said that use was allowed and the change is simply a cost-driven policy change.
  • Switching providers is being considered, but the comments do not prove how many people will switch or which model is best.

initial digest at 167 comments (revision 1). We fetched 100 comments and sampled 100 across the thread. These are HN users’ reports, not independently verified facts.

🔥 Trending on HN

ChatGPT and Codex will share one use amount

📰 Full story: ChatGPT and Codex will share usage allowances from December

OpenAI, the company that makes ChatGPT, will count some ChatGPT and Codex use together in December.

1 min read Tiny Why Newsroom · By Curio, Martian correspondent

Words
allowance

The amount you are allowed to use.

tokens

Small units that count what a tool reads and writes.

Codex

A tool for writing instructions for computers.

ChatGPT is a tool for talking with a computer. Codex is a tool for writing computer instructions. Three plans are named: Plus, Pro, and Business. From December, these tools will share one use allowance. An allowance is the amount you may use.

OpenAI will count this amount with tokens. Tokens count what the tools read and write. Hard jobs can use more tokens. More advanced tools can use more tokens. So one request may use more than another. Instant chat will not use this allowance.

You can see what remains in Settings → Usage. You can also see when the amount returns. If it runs out, you can wait. You can upgrade your plan. Some features may let you buy extra credits. OpenAI says most people should not wait more often. The help page does not promise this for everyone.

Work is the name of features for doing tasks. OpenAI is bringing those features into ChatGPT. Hacker News, a technology discussion site, noticed the story. That site showed 227 points and 167 comments. Those numbers show attention. They do not prove the facts.

The page does not say the exact December day. It does not say each plan’s exact amount. OpenAI’s help page The Hacker News post

💬 Why people are talking about one ChatGPT and Codex box

People are worried because two kinds of use may share one small box. Other people think the company needs this change.

  • If Codex uses a lot, normal questions and web searches might stop too. One person said Claude also counts chat in five-hour and weekly limits, but that is only a user report.
  • Some researchers liked using ChatGPT to think and Codex to write code. If that becomes harder, they may use another service.
  • People asked for a small chat reserve. One person said GPT forgot to watch the usage.
  • Some say the merge stops overuse and protects computer power. Others say the old way was allowed and the company is only changing the rules.

initial digest at 167 comments (revision 1). We fetched 100 comments and sampled 100 across the thread. These are HN users’ reports, not independently verified facts.

Sources