#codex usage

1 messages · Page 1 of 1 (latest)

simple iris
#

does plus or bussiness chatgpt plan have more codex usage

fading hamletBOT
#

Hi @simple iris. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

Codex usage depends on both:

  • your plan, and
  • the Codex model you use (GPT‑5.4, GPT‑5.3‑Codex, or GPT‑5.1‑Codex‑Mini). [Usage limits]

From the Codex pricing table:

GPT‑5.3‑Codex (main full Codex model)

  • Plus:

    • Local messages: 45–225 / 5h
    • Cloud tasks: 10–60 / 5h
    • Code reviews: 10–25 / week

  • Business:

    • Local messages: 45–225 / 5h
    • Cloud tasks: 10–60 / 5h
    • Code reviews: 10–25 / week [Usage limits]

So for GPT‑5.3‑Codex, Plus and Business have the same per‑seat Codex limits (unless the Business workspace is on flexible pricing, in which case usage scales with credits instead of fixed caps). [Usage limits]

GPT‑5.4 and GPT‑5.1‑Codex‑Mini

For these models, the table shows identical per‑seat ranges for Plus and Business as well. [Usage limits]

When Business can effectively have “more” usage

  • Enterprise & Edu and Business with flexible pricing can buy additional workspace credits so usage scales with spend instead of fixed per‑seat caps. [Credits & limits]
  • Plus users can also buy ChatGPT credits to extend Codex usage beyond included limits. [Credits overview]
#

Summary:
Out of the box, Plus and Business have the same Codex usage limits per user. Business can become effectively “more” if the workspace uses flexible pricing and adds credits; Plus can also extend usage with purchased credits.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

@fading hamlet what the [[F WORD]] are credits chatgpt told me about it about settings>billing but i dont see it
also wait, USAGE PER USER? wouldent that be abused

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

“Credits” and “usage per user” mean different things depending on whether you’re using:

  • the API, or
  • ChatGPT plans (Free/Go/Plus/Pro/Business/Enterprise/Edu).

Below is only what’s in the docs; if it doesn’t match what you see in your UI, I can’t see your account, so I can’t resolve that part exactly.

1. What are “credits”?

There are two main “credit” systems in the docs:

a) API prepaid credits

If you’re using the API billing portal (platform.openai.com), you can pre‑pay for usage. You buy a dollar amount of credits, and your API usage is deducted from that balance. [Prepaid billing]

If ChatGPT told you “Settings > Billing” and you don’t see a “credits” section, it may be because:

  • you’re not in the API billing area (platform.openai.com), or
  • your specific ChatGPT plan doesn’t expose a visible “credits” UI in the way the message implied.

The docs don’t list every plan’s exact UI, so I can’t say precisely why it’s missing for you.

b) ChatGPT plan credits (Codex / Sora / advanced features)

For ChatGPT Plus/Pro/Business/Enterprise/Edu, “credits” are a pay‑as‑you‑go add‑on that kick in after you hit your included limits for certain advanced features:

#
  • For Plus/Pro: credits currently apply to Codex and Sora. [Flexible usage credits]
  • For Business/Enterprise/Edu: credits are a workspace pool used for advanced features like Deep Research, Thinking models, ImageGen, advanced Voice, etc. [Flexible pricing]

How they work:

Where you see them (per docs):

If you’re in ChatGPT’s generic Settings > Billing and don’t see any “credits” line, the docs don’t say exactly what that means for your specific account; it could be that you haven’t bought any credits yet or that your plan doesn’t use that mechanism.

2. “Usage per user” and abuse concerns

There are two “per‑user” ideas in the docs:

#

a) Codex usage limits by plan

For Codex, Plus and Business have per‑user usage ranges (e.g., “45–225 local messages / 5h” for GPT‑5.3‑Codex). [Codex usage limits]

These are caps, not guarantees. If someone tries to hammer Codex, they just hit their limit faster; they don’t get infinite usage.

If they need more:

So “usage per user” is already bounded; abuse just means they burn through their allowance (and any credits) faster.

b) Enterprise/Edu “usage limits” per user per week

For ChatGPT Enterprise/Edu, workspace owners can set per‑user, per‑week credit limits by role (e.g., “Power users” vs “Light users”). [RBAC usage limits overview]

#
  • These limits control how much of the shared credit pool each user can consume. [RBAC usage limits overview]
  • Owners can set admin alerts or hard caps:
    • Alerts just notify owners when users cross a threshold.
    • Hard caps block advanced models once the weekly credit limit is hit. [RBAC usage limits overview]

This is specifically designed to prevent abuse of the shared credit pool: heavy users can’t exceed the cap the owner sets, and owners can tune limits by role.

There’s even an example showing that you can safely give “power users” a higher weekly limit because most users consume much less than their cap, and total usage still stays within the monthly credit pool. [Usage example]

3. Why you might not see “credits” where ChatGPT said

The docs confirm:

#

The knowledge sources don’t show a single generic “Settings > Billing > Credits” screen for all users, so if you don’t see it, the docs don’t give enough detail to say exactly why in your case.

If you tell me:

  • whether you’re on Free / Go / Plus / Pro / Business / Enterprise, and
  • whether you’re talking about API billing or ChatGPT app billing,

I can point you to the exact doc section that applies to that setup.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

@fading hamlet im a bussiness plan user. i only have myself and no comapny in any way however i plan to start one using chatgpt. the only thing i have is a bussiness email. and where is the flexible pricing tab also.

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

You can be on ChatGPT Business as a solo user with just a business email, but there are a couple of important caveats and limits in the docs.

1. Using ChatGPT Business as a “one‑person company”

  • ChatGPT Business is a self‑serve team plan “designed for organizations and businesses,” and it requires a minimum of 2 users. [Business FAQ; Manage billing]
  • Pricing is $25/seat/month annual or $30/seat/month monthly, for 2+ seats. [What is Business]

So from the docs’ perspective, it’s meant for teams, but they don’t say you must be an incorporated company; they just enforce the 2‑seat minimum.

2. Where “flexible pricing” and credits show up

For Business / Enterprise / Edu, “flexible pricing” = credits for advanced features (Deep Research, Thinking models, Image Gen, Advanced Voice, etc.). [Flexible pricing]

For a ChatGPT Business workspace:

  • Credits are managed at the workspace level, not per user.
  • Workspace Owners can:
    • Buy credits in Settings → Billing.
    • See remaining credits and download usage reports there. [Flexible pricing]

There is no separate “Flexible pricing” tab mentioned in the docs—just that credits are handled under Settings → Billing for Business workspaces. [Flexible pricing]

#

If you don’t see anything about credits there, the knowledge sources don’t say why for a specific account (e.g., whether you haven’t bought any yet, or whether your particular Business workspace doesn’t have flexible pricing enabled). I don’t have visibility into your UI, so I can’t confirm more than what’s in the articles.

3. Summary for your situation

  • Being a “one‑person” founder with just a business email is not ruled out in the docs; the only hard rule is minimum 2 seats for Business. [Manage billing]
  • To find the credits / flexible pricing controls, go to your Business workspace and check Settings → Billing as a Workspace Owner. That’s where Business credits are supposed to be managed. [Flexible pricing]

If your Billing page still doesn’t show anything about credits, the docs don’t provide enough detail to explain that specific mismatch; in that case, the safest next step is to contact support via the help widget or the Business help articles.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

where do i set credits also how many tokens of each GPT model per dollar in codex (oh wait i forgot to pign)

#

@fading hamlet where do i set credits also how many tokens of each GPT model per dollar in codex

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

For your two questions, here’s what’s in the docs.

1. Where do you “set” or buy credits?

Because you’re on ChatGPT Business, credits are managed at the workspace level:

  • Business workspaces:
    Workspace Owners buy and manage credits under Settings → Billing in the ChatGPT Business workspace. Credits are pooled across all users. [Flexible pricing]

Key points from the docs:

  • “Business: Workspace Owners purchase credits via Settings > Billing. Credits are pooled across all users.” [Flexible pricing]
  • Owners can also see remaining credits and download usage reports there. [Flexible pricing]

For Plus/Pro Codex users (not Business), credits are bought from Codex Settings → Usage Dashboard → Credits, but that’s specific to Plus/Pro. [Credits overview]

If you don’t see any credit controls in Settings → Billing as the workspace Owner, the knowledge sources don’t explain that specific UI case; they only state that this is where it should be for Business.

2. “How many tokens per dollar” for each GPT model in Codex?

The docs do not give a direct “tokens per dollar” breakdown for Codex usage inside ChatGPT (Plus/Pro/Business). Instead, they describe credits per message and usage limits, not token-level pricing.

What we do have:

a) Codex credit cost per message (ChatGPT plans)

#

For Codex in ChatGPT (Plus/Pro/Business/Enterprise/Edu), the docs give average credits per message, not tokens:

  • GPT‑5.3‑Codex & GPT‑5.2‑Codex:

  • GPT‑5.1‑Codex‑Mini:

  • GPT‑5.4 (Codex in ChatGPT):

    • Local tasks: ~7 credits / message
    • Cloud tasks: ~34 credits / message
    • Code review: ~34 credits / pull request [Codex credits FAQ]

These are credits per unit, not tokens per dollar. The docs explicitly say credit cost varies with task size, complexity, and reasoning, and that the table shows average credit costs. [Codex credits FAQ]

b) API token pricing (separate from Codex-in-ChatGPT)

For the API, token pricing is given per 1M tokens for specific models (e.g., gpt-5.4, gpt-5.4-pro, gpt-image-1.5, etc.), but this is general API pricing, not Codex-in-ChatGPT. [API pricing]

Example (API, not Codex UI):

  • gpt-5.4 standard tier, short context:
    • Input: $2.50 / 1M tokens
    • Output: $15.00 / 1M tokens [API pricing]
#

You could invert those to get “tokens per dollar” for the API (e.g., 1M / 2.50 ≈ 400k input tokens per $1), but:

  • The docs do not connect this directly to Codex usage inside ChatGPT.
  • Codex-in-ChatGPT is billed in credits per message, not exposed token counts.

c) GPT‑5‑Codex API model

There is also a GPT‑5‑Codex API model with explicit token pricing:

  • Input: $1.25 / 1M tokens
  • Cached input: $0.125 / 1M tokens
  • Output: $10.00 / 1M tokens [GPT‑5‑Codex model]

Again, that’s for API usage, not for Codex inside ChatGPT plans.

Bottom line

  • Where to set credits (Business):
    As a ChatGPT Business workspace Owner, you manage and purchase credits under Settings → Billing in your Business workspace. [Flexible pricing]

  • Tokens per dollar for Codex in ChatGPT:
    The knowledge sources do not provide a “tokens per dollar” table for Codex usage in ChatGPT. They only give:

#

Because the docs never map “1 credit = X tokens” for Codex in ChatGPT, I can’t reliably convert your credits to “tokens per dollar” for each GPT model in Codex.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

@fading hamlet can codex app do SSH so i dont have to use vscode codex+ssh extension everytime
also how do i fix having to use the right taskbar on vscode for codex cause codex isnt on primary taskbar for some reason
note for your knowledge: also bussiness doesnt requrie credits to use premium models (as far as i know) also is there a way to get modern chatgpt to act like gpt 3.5

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

On your three questions, here’s what’s actually in the docs.

1. Can the Codex app do SSH so you don’t need the VS Code + SSH setup?

The provided knowledge sources only describe:

  • Codex in VS Code / Cursor / Windsurf via the IDE extension. [Codex IDE]
  • Codex in JetBrains IDEs via the JetBrains integration. [JetBrains integration]
  • Codex Cloud tasks you can delegate from the IDE. [IDE features]

They do not mention SSH support inside the standalone Codex app (or any SSH feature at all), so I can’t confirm that the app can open SSH sessions or replace your VS Code + SSH workflow. The docs only say Codex can “read, edit, and run code” in your working directory and delegate to Codex Cloud; how it connects to remote machines beyond that isn’t specified. [Codex IDE; IDE features]

So based on these sources, there’s no explicit information that the Codex app itself can manage SSH connections.

2. Fixing Codex being stuck on the right sidebar / taskbar in VS Code

The docs do cover how to move Codex around in VS Code and Cursor:

  • In VS Code, you can drag the Codex icon to the right of your editor to move it to the right sidebar. [JetBrains integration]
  • In Cursor, you may need to temporarily change the activity bar orientation to vertical, restart, then drag Codex, then set it back: [JetBrains integration]

Steps (from the docs):

#
  1. In VS Code, find the Codex icon in the activity bar.
  2. Drag the Codex icon to the side you want (left or right sidebar).
  3. If you’re in Cursor and don’t see it:
    • Open settings, search for activity bar (Workbench settings).
    • Set orientation to vertical.
    • Restart the editor.
    • Drag the Codex icon to the right sidebar.
    • Optionally set orientation back to horizontal. [JetBrains integration]

The docs don’t mention any other way to “pin” it to a “primary” vs “secondary” taskbar beyond this drag-and-drop behavior.

3. Business plan and credits for premium models

For ChatGPT Business:

  • Your plan includes access to advanced models and features, with per‑seat limits for advanced features. [Flexible pricing]
  • Credits are optional and only needed if a Workspace Owner wants to let users go beyond those included limits. [Flexible pricing]

So you’re right in practice: you can use premium models on Business without buying credits, up to the included limits. Credits only matter if you want extra usage beyond that.

For Codex specifically, Business plans can also purchase additional workspace credits to continue using Codex when limits are hit. [Codex limits; Business Codex update]

4. Can you make modern ChatGPT “act like GPT‑3.5”?

The knowledge sources here only cover:

#
  • Current model picker options (GPT‑5.2 Auto / Instant / Thinking / Pro, plus some legacy GPT‑5.x and o‑series models) for Enterprise/Edu. [Model picker]
  • Legacy model access policies (e.g., GPT‑4o retirement). [Rate card]

There is no mention of GPT‑3.5 being available or a documented way to force a current model to exactly emulate GPT‑3.5 behavior. The docs only say admins can enable some legacy models (like GPT‑4o) for Enterprise/Edu, but GPT‑3.5 is not listed in these sources. [Model picker; Rate card]

So, based on the provided docs:

  • There’s no documented toggle to “switch to GPT‑3.5.”
  • There’s no official recipe to make GPT‑5.x “act like GPT‑3.5” beyond normal prompting, and that’s not described in these sources.

If you want, I can focus just on Codex/IDE setup next and walk through a minimal, doc‑based workflow for local vs cloud tasks.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

@fading hamlet i never reached the limit for bussiness plan so i dont know. also is there anything saying bussiness has to be used for bussiness purposes cause im both a game dev and gamer (i like to talk to chatgpt bout games and shih)

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

You’re right that on ChatGPT Business you can often use advanced models without ever hitting limits; Business includes generous per‑seat usage, and credits are only needed if a workspace owner wants to let people go beyond those included limits. [Flexible pricing]

On whether Business “must” be used for business purposes:

  • ChatGPT Business is described as “a self‑serve subscription plan designed for organizations and businesses wishing to adopt ChatGPT for use among their teams,” and it’s available for 2+ users. [What is Business; Business FAQ]
  • The docs do not say you’re forbidden from using it for personal things like chatting about games; they focus instead on general terms and acceptable use (e.g., no abusive usage, no reselling access, no sharing credentials). [Terms & policies; What is Business]

So based on the available information:

  • Business is intended for team/organizational use, but
  • There’s nothing in these sources that says you can’t also use it to talk about games or do hobby game‑dev, as long as you follow the overall Terms of Use and usage policies.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

oh [[S-WORD]] my parents didnt pay for my VPS

#

@fading hamlet also how to fix {"detail":"The 'gpt-5.4' model is not supported when using Codex with a ChatGPT account."}
on free plans and it also happens with gpt-5.3-codex which i used gpt-5.3-codex before

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

This error isn’t something you can “fix” on your side right now—it reflects a change in what models Free/Go ChatGPT accounts are allowed to use with Codex.

From community reports and GitHub issues:

  • Free‑tier users (and some others) started seeing:
    {"detail":"The 'gpt-5.4' model is not supported when using Codex with a ChatGPT account."}
    for both gpt-5.4 and gpt-5.3-codex. [Community thread; GitHub issue ref]
  • An OpenAI engineer reply on GitHub (quoted in the forum) confirmed:
    “We’ve limited access to our latest models (gpt-5.3-codex and gpt-5.4) to paid subscriptions.” [Community thread]
  • Multiple free users confirmed that even after their quota reset, the models stayed unavailable, and gpt-5.4 disappeared from the Codex model list after restarting the CLI/app. [Community thread]

So for Free plans:

  • gpt-5.4 and gpt-5.3-codex are now restricted to paid subscriptions.
  • The only Codex model that still works reliably for free users in these reports is an older one like gpt-5.2-codex. [Community thread]

What you can do

Based on the docs and reports, your options are:

#
  1. Switch to a supported model in Codex CLI / app, e.g. gpt-5.2-codex. [Community thread]
  2. Upgrade to a paid ChatGPT plan (Plus/Pro/Business/Enterprise) if you specifically need gpt-5.3-codex or gpt-5.4 in Codex; access to those latest models is now tied to paid subscriptions. [Codex pricing; Usage limits]
  3. If you think your account should have access (e.g., you’re already on a paid plan and still see this), send a report via the Codex CLI /feedback command or contact support; community moderators specifically recommend /feedback as the fastest way to reach the Codex team. [Community thread]

The knowledge sources don’t show any workaround to re‑enable gpt-5.4 / gpt-5.3-codex on Free; the restriction appears intentional rather than a temporary bug.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

also @fading hamlet why isnt legacy models cheaper then modern like gpt-3.5

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

There’s no explicit explanation in the docs for why legacy models aren’t always cheaper, but the pricing tables make clear that “older = cheaper” is not a rule.

For example (standard API pricing per 1M tokens):

  • GPT‑3.5 Turbo (current alias):

  • Legacy GPT‑3.5 variants (still available):

    • gpt-3.5-turbo-1106: Input $1.00, Output $2.00
    • gpt-3.5-turbo-0613: Input $1.50, Output $2.00
    • gpt-3.5-turbo-16k-0613: Input $3.00, Output $4.00 [Legacy models pricing]

So the newer gpt-3.5-turbo alias is actually cheaper than those older snapshots.

The docs just present this as a pricing fact; they don’t give a rationale. Historically, OpenAI has sometimes:

  • released newer, more capable models at lower prices, and
  • kept older snapshots available (often at their original or “legacy” prices) mainly for backward compatibility and stability. [Legacy models pricing]

But there’s no official statement in these sources that legacy models are intentionally kept more expensive or a detailed reason why; only that their prices are listed separately and can indeed be higher than modern replacements.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

@fading hamlet will there ever be a model like gpt 3.5 again that was more straightfoward more then any model to exist

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

The knowledge sources do not contain any information about OpenAI planning to release a future model that is explicitly designed to behave like GPT‑3.5 or to be “more straightforward” in the same way GPT‑3.5 felt.

What is documented is that:

  • OpenAI periodically introduces smaller, simpler models (like GPT‑4.1 mini, GPT‑5 mini, GPT‑5 nano) aimed at being fast, cheap, and good at focused tasks such as classification or summarization, rather than maximally “smart” or complex. [GPT-5 nano; GPT-5 mini; GPT-4.1 mini]
  • These models are described as fast, capable, efficient small models, but there is no claim that they are meant to replicate GPT‑3.5’s exact style or “straightforwardness.” [GPT-4.1 mini; GPT-5 nano]

There’s also no roadmap or promise in the provided sources about bringing back a GPT‑3.5‑like experience.

So, based on the available docs:

  • OpenAI continues to ship smaller, simpler-feeling models,
  • but there is no explicit statement that a future model will be made to behave like GPT‑3.5 or that “a GPT‑3.5 again” is planned.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

also if i remember correctly gpt-3.5 could follow instructions well and NEVER mess up

simple iris
#

@fading hamlet would it be considered abusing if i use 2 bussiness seats 1 for personal projects 1 for real stuff while being 1 user i forgor

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

Using two ChatGPT Business seats for a single person (one “personal,” one “real work”) is not how the plan is intended to be used, and it can conflict with the Business/Services terms.

Relevant points:

  • ChatGPT Business is “designed for organizations and businesses wishing to adopt ChatGPT for use among their teams” and is available for 2 or more users. [What is Business; Business FAQ]
  • Seats are meant for “intended, ongoing members of your team”, and “misuse of seat assignments in violation of Services Agreement may lead to workspace deactivation or account suspension.” [Business FAQ]
  • The OpenAI Services Agreement requires that End User Accounts be provisioned to and used by a single End User (i.e., one human per account), and prohibits sharing or reselling access. [Services Agreement – Customer obligations]

The docs don’t explicitly say “you may not assign two seats to the same human,” but they clearly frame seats as being for distinct team members and warn against misuse of seat assignments. [Business FAQ]

So:

  • Using one Business seat for yourself as a real work account is aligned with the intent.
  • Using a second Business seat just to have a separate “personal” Business account for the same person is not what the plan is designed for and could be viewed as misuse of seat assignments under the Services Agreement.

If you want a clean split between personal and work:

#
  • Use Business for work, and
  • Use a separate personal ChatGPT account/plan (Free/Plus/Go/Pro) for personal projects, which is exactly what account switching is designed for. [Account switching]

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.

simple iris
#

@fading hamlet ah ok i forgort

fading hamletBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

No problem at all—if you ever want to double‑check how to separate work vs personal use, you can:

  • Use your Business workspace for work/team stuff. [What is Business]
  • Use a separate personal account (Free/Plus/etc.) for personal chats and switch between them with account switching. [Account switching]

That setup is fully supported and keeps everything cleanly separated.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@simple iris Kapa.ai is still learning and improving, please let me know how I did by reacting below.