Please offer a fixed-price high-usage individual plan for Codex and agent features (effectively unlimited under fair use)

[buraburachinpokou]

I would like to request a fixed-price, high-usage individual subscription tier for Codex and other agent features.

The main issue with the current usage-limited model is not simply that limits exist.

For users doing serious development or long multi-step agent work, constantly thinking about remaining capacity changes how we use the product:

• we hesitate to run necessary tests,
• we avoid useful iterations,
• we delay experiments,
• we may stop development even when the task itself is worth continuing,
• and unexpected repair/debugging loops can consume a large amount of capacity through no fault of the user's original plan.

This makes usage limits part of the development decision itself.

What I would like

A personal subscription tier with:

• a fixed monthly price,
• a very high usage allowance for Codex / agent work,
• no need to repeatedly purchase extra credits during normal heavy personal use,
• and fair-use restrictions only for extreme or abusive workloads.

It does not have to be literally unlimited.

An “effectively unlimited for normal individual heavy usage” model would solve the problem.

Example price point

Personally, I currently use ChatGPT Plus, but I would be willing to pay around JPY 10,000/month for a plan that gives me enough Codex / agent capacity that I do not need to constantly monitor usage.

The important part is predictability:

fixed monthly cost → use the agent when needed → no repeated capacity decisions

rather than:

subscription → hit capacity → decide whether to buy more credits → repeat

Why this matters for development workflows

Agentic coding is often unpredictable.

A task that looks small can turn into:

• implementation,
• test failure,
• debugging,
• regression testing,
• another fix,
• another validation pass.

When every additional iteration consumes a visibly limited pool, users can become reluctant to let the agent finish the job properly.

That creates the wrong incentive.

For development work, I would rather pay a higher fixed monthly fee and optimize for:

quality of result

instead of:

minimizing agent usage because I am afraid of exhausting capacity.

Suggested plan structure

For example:

High-Usage Individual / Developer Plan

• Fixed monthly subscription
• Much higher Codex and agent usage than Plus
• Effectively unlimited for normal individual development under fair use
• Clear protection against abuse / automated industrial-scale usage
• No repeated manual credit purchases for ordinary heavy usage
• Same ChatGPT account, context, memory, and tools

This would fill the gap between casual Plus usage and unpredictable pay-as-you-go expansion.

Why I would pay for it

The value of an AI coding agent increases when I can use it freely enough to:

• experiment,
• test,
• retry,
• debug,
• and validate thoroughly.

If I have to constantly calculate whether a task is “worth spending capacity on,” the product becomes less useful even before the limit is actually reached.

I would prefer a more expensive but predictable subscription that lets me focus on the work instead of the remaining quota.

For my own use, a plan around JPY 10,000/month would be reasonable if the practical usage allowance were high enough that capacity stopped being a normal concern.

[JPM-concordia-labs]

Have you tried PRO subscription, the 5x version (100$/Month) or even the 20x versions (200$/Month).
I do not see why would OpenAI add another tier for 50$ between Plus and Prod 5x.

[hoodwinky]

Because 3.5 million (1 in 3) people in Australia are experiencing food poverty? Because the world is already designed to advantage the rich and make the poor poorer?

[hoodwinky]

I'm inclined to agree. Even the $20 add on is way too expensive ($32 AUD, as the prices are in USD) and hardly gives you any more tokens to truly work with. For those of us on a low income with disabilities, and in my case are also studying, who are learning and using tools to help improve our situations and prospects, it is definitely priced in a way that significantly disadvantages you based on your income-level. Most low to middle income earners and students do not have $100+ a month to spend on ChatGPT. A policy around usage for students and people experiencing financial hardship would be amazing! Verify students like other sites that offer low priced access (check they receive emails to an educational email address), and manual processing for those in severe financial hardship who could benefit from the access. The way it is at the moment: ChatGPT and advancing yourself with the use of LLMs and AI is starting to go very much towards being a rich people perk, and less about being the "OPEN" AI you'd have hoped for....

[JPM-concordia-labs]

Student subscription is something i've seen done https://chatgpt.com/students/2026/ I believe is US only for now but i would not be surprised if it expands to other geographies.
Regarding special subscriptions for people in financial hardships, I do not see this feasible. First - this would mean that to open up a subscription you would need not only a valid email and couple personal data fields but also your Tax returns, maybe some government documents as proof of financial need - is this information we want to be sent to a company? I even see all the GDPR issues regarding this for EU countries. Secondly, this is like saying that a landlord has the obligation to ask for lower rents if a person has low income; sounds good in theory but, in reality, this is making private businesses/People do a job that is of the government: Social Security / welfare.
Instead of asking companies, like OpenAI, Anthropic, Google, etc.., to create plans for specific groups of people, why not ask local schools, even government, to build a partnership with said companies for better ratings?
Microsoft, for example, has this is colleges. Colleges would enter into a partnership with Microsoft so its students could enroll and have access to a specific student tier of Office, Azure, and other apps.
Also, on the first sentence you said " Because 3.5 million (1 in 3) people in Australia are experiencing food poverty?" ---> i believe that people in a situation that they need food, having AI access - even free, its not on their top 10 priorities. Maybe they need their welfare issues solved first? Someone that struggles to buy food should not be spending 10, 20 or 30$ on subscriptions (AI, netflix, and so on).

[hoodwinky]

Sorry, you’re absolutely right. Those weren’t the best examples. Let me reframe what I meant. If someone is in need and can identify an opportunity to use AI to improve their situation, or the situation for many, then giving them access for a limited time period wouldn’t hurt. There’s no need for requesting information or linking to personal info. Just an identified need and an idea to solve that need/project that’s obviously going to be identifiable based on the fact they ran out of usage to begin with. If you’re having a hard time and you think you truly think you can benefit from the help and have run out of data? Take what you need, within limits - they’re using their brain to find a way out. Psychologically you wouldn’t want to suddenly place a barrier between that person in distress and a perceived solution to their problem. Why knock a person down as they’re working their way up? Maybe it’s a naive and toxically positive view that I’m taking. But, as you said, how many people would that really be of use to? Worthy enough amount of people to give a hall pass to for a specified amount of time. There would be a limit. But I’m sure it wouldn’t be that hard to identify the person who became frustrated when the limit was reached and who was truly working on something productive and non-commercial. Go for it, I’d say anyway. In debt and need to brainstorm your way out of it… Why not? Waiting lists for support services are crazy. Sometimes it’s easier to help yourself with a nudge from a friend (even if that friend is ChatGPT) Protective in a way, preventing a person from having a potential breakdown caused by something they were working so hard on in order to achieve their way out of a debilitating problem… If it’s you and you’re reading this- I see you 👀 And a definite yes please to rolling out benefits for students outside the US! We need not lose our compassion for one another for it is that that makes us human. Design for the user, not for the company bottom line.

…

-------- Original Message --------

On Wednesday, 08/26/26 at 00:48 Concordia_AI_labs ***@***.***> wrote: Student subscription is something i've seen done https://chatgpt.com/students/2026/ I believe is US only for now but i would not be surprised if it expands to other geographies. Regarding special subscriptions for people in financial hardships, I do not see this feasible. First - this would mean that to open up a subscription you would need not only a valid email and couple personal data fields but also your Tax returns, maybe some government documents as proof of financial need - is this information we want to be sent to a company? I even see all the GDPR issues regarding this for EU countries. Secondly, this is like saying that a landlord has the obligation to ask for lower rents if a person has low income; sounds good in theory but, in reality, this is making private businesses/People do a job that is of the government: Social Security / welfare. Instead of asking companies, like OpenAI, Anthropic, Google, etc.., to create plans for specific groups of people, why not ask local schools, even government, to build a partnership with said companies for better ratings? Microsoft, for example, has this is colleges. Colleges would enter into a partnership with Microsoft so its students could enroll and have access to a specific student tier of Office, Azure, and other apps. Also, on the first sentence you said " Because 3.5 million (1 in 3) people in Australia are experiencing food poverty?" ---> i believe that people in a situation that they need food, having AI access - even free, its not on their top 10 priorities. Maybe they need their welfare issues solved first? Someone that struggles to buy food should not be spending 10, 20 or 30$ on subscriptions (AI, netflix, and so on). — Reply to this email directly, [view it on GitHub](#40291?email_source=notifications&email_token=CMLKD2RHA54L6SAGTHRBLRL5LWRLLA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBRGQ4TQNBUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-18149844), or [unsubscribe](https://github.com/notifications/unsubscribe-auth/CMLKD2R54Q2OURQAAPIAUNL5LWRLLAVCNFSNUABIKJSXA33TNF2G64TZHM4TMNJUGE2TMNBZHNCGS43DOVZXG2LPNY5TCMBWG4ZTANRWUF3AE). Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for [iOS](https://github.com/notifications/mobile/ios/CMLKD2UQ74N7RRMLUOGR7I35LWRLLA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBRGQ4TQNBUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSVGM33PORSXEX3JN5ZQ) and [Android](https://github.com/notifications/mobile/android/CMLKD2Q3M5FSXPBJFN2SSMT5LWRLLA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBRGQ4TQNBUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ). Download it today! You are receiving this because you commented.Message ID: ***@***.***>

[bTn138]

I’d love to see a separate Developer plan focused on Codex and agent usage something between Plus and Pro. with higher Codex limits and better token efficiency

Id also like more control over optimization, like automatically using cheaper models for simple tasks and stronger models only when needed, so agents and tokens are used as efficiently as possible