Categories
Artificial Intelligence

ChatGPT Plus and Codex: The Usage Limits the Pricing Page Doesn’t Make Obvious

So simply ask over ChatGPT.

Yes — your concern is reasonable. I checked the current OpenAI documentation and pricing pages, and there is an important distinction between “Codex is included with Plus” and how much Codex usage Plus actually includes.

What Plus currently gives you

The main ChatGPT pricing page describes Plus as having “Expanded Codex usage”, but it does not show a simple numeric Codex allowance directly beside the Plus purchase option. It links to a separate “Limits apply” page. citeturn1view0

Codex/Work usage is metered separately from ordinary ChatGPT chatting. OpenAI currently says that usage can be constrained by both a 5-hour window and a weekly window, and you need allowance remaining in the applicable limits to continue. citeturn1search6turn1search1

More importantly, it isn’t simply:

“You get X Codex prompts per week.”

Consumption varies according to the model, reasoning level, task complexity, context size, tools, and how long the task runs. A large repository or long agentic coding task can therefore consume dramatically more allowance than a small coding request. citeturn0search0

That is probably why it feels like the allowance disappears surprisingly quickly when you’re doing substantial development work.

Why wasn’t this obvious when buying Plus?

This is the part where the current presentation isn’t especially transparent.

The purchase page currently says:

Plus → “Expanded Codex usage”

and then:

“Limits apply”. citeturn1view0

It does not, in the main Plus feature list, say something straightforward such as:

Codex has a separate usage allowance.
Heavy coding tasks consume more allowance.
A weekly limit applies.
Work and Codex may share an allowance.

You have to follow the documentation to discover those details.

The Help Center is considerably clearer: it explicitly says that Codex, ChatGPT Work and some other agentic features can draw from the same allowance/credit pool, and that large codebases, extended sessions and long-running tasks consume substantially more. citeturn0search0turn0search8

So if your impression when purchasing Plus was “Plus includes Codex, therefore I’ll be able to use Codex reasonably freely throughout the month,” the pricing-page wording alone doesn’t give you a particularly good idea of how restrictive heavy Codex use may become.

One other thing worth knowing

Once the included allowance is exhausted, eligible Plus users can purchase additional credits rather than necessarily upgrading the whole subscription. OpenAI says the included allowance is consumed first and then eligible usage can draw from purchased credits. citeturn0search8

You can check the actual state of your account under Settings → Usage. Codex CLI also supports /status; OpenAI specifically recommends checking the usage dashboard for the exhausted allowance and reset time. citeturn0search0

urlOpenAI’s current ChatGPT pricing pageturn1view0
urlOpenAI’s Codex usage explanationturn0search0

Categories
AI Humor Artificial Intelligence Digital Life Technology,

When AI Took Me Too Literally: A Collection of Funny AI User Incidents

Artificial Intelligence is getting smarter every day. It can write emails, generate images, summarize books, create websites, and even help students with homework. But sometimes, the funniest moments happen when humans and AI misunderstand each other in spectacular ways.

Here are some real-life-inspired incidents that show why AI still has a long way to go before it truly understands human beings.


1. “Make My Logo Bigger”

A business owner uploaded a banner design and told the AI:

“Can you make the logo slightly bigger?”

The AI responded by increasing the logo size from 10% of the banner to approximately 90%.

The logo became the banner.

The business name disappeared.

The contact details disappeared.

The slogan disappeared.

The only thing visible was a giant logo staring confidently at everyone.

Technically, the request was fulfilled.


2. The Resume Disaster

A job seeker asked AI:

“Make my resume stand out.”

The AI enthusiastically rewrote his experience.

Original:

Worked at XYZ Company for 2 years.

AI Version:

Visionary Technology Leader Transforming Enterprise Digital Ecosystems Through Strategic Innovation.

The candidate was actually a junior support executive resetting passwords.

The interviewer spent 15 minutes trying to understand how a “Visionary Technology Leader” forgot his own email password.


3. The Angry Customer Email

A frustrated customer wrote:

Write a polite email saying I am unhappy.

AI generated:

Dear Sir,

I hope this email finds you well. I would like to express my deepest gratitude for the unique opportunity to experience disappointment at a level I previously believed impossible.

The customer laughed so hard he forgot to be angry.


4. The Student’s Shortcut

A student asked:

Solve this math problem and explain simply.

AI provided a detailed explanation.

The student replied:

Simpler.

AI simplified it.

Student:

Simpler.

AI simplified again.

Student:

Simpler.

Finally AI responded:

Number go up. Answer 42.

The student submitted it.

The teacher wrote:

Marks go down.


5. The Restaurant Menu Incident

A restaurant owner wanted fancy menu descriptions.

Original:

French Fries

AI Version:

Handcrafted golden potato batons delicately crisped to perfection and accompanied by an immersive flavor experience.

Customers expected a luxury dish.

They received fries.

Good fries, but still fries.


6. The Overly Honest AI

A user asked:

Am I productive today?

The AI reviewed the user’s activities:

  • Opened 17 tabs
  • Watched 43 productivity videos
  • Reorganized desktop folders
  • Renamed files
  • Read articles about productivity

Then replied:

You have spent approximately 8 hours preparing to become productive.

That one hurt.


7. The Website Launch

A developer asked AI:

Create a modern website.

AI generated:

  • Dark theme
  • Animations
  • Gradients
  • Glassmorphism
  • Shadows
  • More shadows
  • Even more shadows

The page looked beautiful.

Loading time: 18 seconds.

The developer proudly launched it.

Visitors admired the loading spinner.


8. The Diet Plan

User:

Give me a healthy diet plan.

AI:

Eat vegetables, fruits, proteins, and drink water.

User:

I don’t like vegetables.

AI adjusted.

User:

I don’t like fruits.

AI adjusted.

User:

I don’t like protein.

AI paused.

Then responded:

Have you considered photosynthesis?


9. The Smart Home Problem

A user connected AI to their smart home.

User:

Make the room cozy.

AI interpreted:

  • Lights: 20%
  • Temperature: 22°C
  • Soft music
  • Curtains closed

Perfect.

Then the user said:

A little more cozy.

AI turned off all lights.

Nobody could find the switch.


10. The Ultimate AI Question

One evening a user asked:

Can you replace humans?

AI replied:

Who would ask me questions if I did?

For a brief moment, both human and machine agreed on something.


The Real Lesson

Most funny AI stories happen because humans assume AI understands context the same way people do.

Humans communicate with hints, emotions, assumptions, and incomplete sentences.

AI communicates with patterns, probabilities, and sometimes an alarming level of literal interpretation.

The result?

Occasional confusion.

Unexpected comedy.

And stories worth telling.

As AI becomes more powerful, one thing remains certain:

The funniest bugs in technology will always come from the interaction between human creativity and machine logic.

And honestly, we wouldn’t want it any other way.