You’ve probably seen a post claiming ChatGPT has “secret codes,” the same kind of framing you’ll see for Samsung’s dialer codes or OnePlus’s hidden menus. This guide looks at the real story behind ChatGPT prompt codes for students: which ones actually do something, which are just custom instructions dressed up as commands, and where the one genuinely official feature, Study Mode, fits in. Type ELI5 and it explains things simply. Type KILLCRITIC and it stops flattering you. It sounds like you’re unlocking a hidden setting, and that’s exactly why the framing is a little misleading.
These are not toggle switches in ChatGPT. Go into the tools menu and see that the KILLCRITIC option is nowhere to be found. What’s actually going on here is a lot more straightforward, and ultimately more helpful when you know what’s going on you’re training the model how to respond to an instruction you made up yourself.
There is one real exception worth knowing, though, because people mix it up constantly with the fake codes: Study Mode. That one’s an actual OpenAI feature with documented activation steps, not a keyword someone made up for a TikTok.
Are ChatGPT prompt codes for students actually real?
It depends on what you consider “real.” It is not hardcoded in ChatGPT as Study Mode is, but it is not false either because these prompts can actually influence ChatGPT’s responses since you programmed it to act differently upon recognizing this word. Think of it less like a command and more like a nickname you give to a longer instruction so you don’t have to retype it every time. That’s really all ChatGPT prompt codes for students are: nicknames for instructions you wrote.
How to set up your own ChatGPT prompt codes
From now on, whenever I type ELI5, explain the topic I give you as if I
were five years old. Use a simple comparison and no technical terms.
Send that once, and “ELI5: mitosis” works for the rest of that chat. It usually holds up reasonably well within a single conversation, though the interpretation can drift a bit if the thread runs long. Want it to survive into new chats too? Drop the definition into Custom Instructions under Settings. It’ll carry over more often than not, but don’t expect identical wording every single time. If you’ve got Memory turned on, ChatGPT might also pull in things it’s picked up from earlier conversations, which is a separate mechanism from the code itself, worth knowing but not something to lean on.
The best ChatGPT prompt codes for students, worth actually using
ELI5 and ELI10:- Explain it like I’m five, or like I’m ten. The five version is good for a first pass at something genuinely confusing; the ten version works better once you already grasp the basics and just want the gap filled in without a lecture.
ELI5: [topic]
Explain this using a simple, everyday comparison.
Then check I understood it by asking me one question.

KILLCRITIC:- This is the one that people are talking about most, and it’s understandable why. ChatGPT can be too nice when asked to evaluate someone’s writing, which can lead to people not addressing their weaker arguments and poorly written paragraphs before submitting their work. KILLCRITIC lays out what needs to be said more clearly, telling the model to be brutal about the bad writing and arguments while avoiding the unnecessary praise that ChatGPT sometimes adds in.
From now on, when I type KILLCRITIC before a piece of writing, give me
direct, critical feedback only. Point out weak arguments, unclear
sentences, and factual errors. Do not soften the tone or add compliments.

X10THINK. Nudges the model to slow down on a multi step problem instead of racing to the first plausible answer. It helps, but it’s not magic. A more careful looking answer isn’t automatically a correct one, so check the arithmetic yourself before you write it on the exam.
X10THINK: [your problem]
Solve this carefully in clear, checkable steps. State the assumptions,
show the key calculations, verify the result using a second method when
possible, and flag any uncertainty.
ALT3. Three genuinely different angles on the same request, useful when your first idea for an essay outline or a project just isn’t clicking.
ALT3: [your request]
Give me three genuinely different approaches, not three versions of the
same idea. Briefly note the trade offs of each.
HINTS. Stops it from just handing over the final answer.
Help me solve this problem without giving me the final answer right
away. Give me one hint at a time and wait for my attempt before
continuing. Problem: [paste your problem here]
QUIZME. Turns your own notes into a back and forth quiz instead of a wall of text you skim once and forget.
QUIZME
Use the notes below to quiz me one question at a time. Wait for my
answer each time. Grade it, explain what I got wrong, and then ask a
slightly harder question. Notes: [paste here]
EDITMYDRAFT. For polishing your own writing, the same job a lot of free AI writing tools are built around. Not for making generated text pass as entirely yours where that breaks your school’s rules; those are different uses with very different consequences.
EDITMYDRAFT: [paste your own draft]
Improve clarity, flow, grammar, and sentence variety without changing
my meaning or adding ideas. Mark major changes and explain what you changed.
FLASHCARDS. Fast, but only as good as what you feed it. Skim the output against your actual notes before trusting it near an exam.
FLASHCARDS: [paste your notes]
Turn these notes into flashcards, one fact or definition per card.
Format each as Front: / Back:. Keep answers short enough to recall
quickly. Flag anything you're unsure about instead of guessing.
SUMMARIZE. For when a reading is long and your evening isn’t.
SUMMARIZE: [paste text or describe the source]
Give me a summary in [number] bullet points, ordered by importance.
Then list two things a professor would likely test me on.
COMPARE. Two theories, two events, two characters, whatever your revision keeps blurring together.
COMPARE: [item one] vs [item two]
Lay out their similarities, their differences, and one common mistake
students make when confusing them.
MNEMONIC. For the list you can recognize but can never recall in order under pressure.
MNEMONIC: [list of terms or steps]
Create a memorable mnemonic or short phrase to help me recall this list
in order. Explain how each part of the mnemonic maps to the original item.
STUDYPLAN. Built around the time you actually have left, not a generic calendar template.
STUDYPLAN
I have [number] days before my exam on [subject]. My weakest topics are
[list them]. Build a day by day plan that spends more time on my weak
areas, includes short breaks, and ends with a light review day.
Building a diagram with ChatGPT prompt codes
Typing DIAGRAM doesn’t make ChatGPT draw anything by itself, it’s just another custom code, and it’s worth being upfront that “diagram” means different things depending on what you actually want. This is one of the more requested ChatGPT prompt codes for students who need to visualize a process rather than just read about it.
If you just need the content organized so you can sketch it yourself, ask for a structured description:
DIAGRAM: [topic]
Describe this as a labeled diagram. List each stage or part in order,
what happens at each one, and how the arrows or connections between
them should be drawn.
If you want something that actually renders as a chart, ask for Mermaid syntax instead. You won’t see the picture inside the chat, but you can paste the code into a free Mermaid live editor or plenty of note apps and it draws itself.
Write a Mermaid flowchart showing the stages of [process or topic],
with short labels at each step. Use flowchart TD syntax so it renders
top to bottom.
And if your plan includes image generation, you can ask ChatGPT to generate an actual picture of a diagram right in the chat. That’s a genuine built in feature, not a code you’re defining, and it’s decent for something illustrative. Don’t count on it for anything needing precise labels, since text inside generated images tends to come out mangled.
If you’re weighing ChatGPT against other assistants for tasks like this, it’s worth understanding the actual difference between an AI agent and a plain chatbot first, since that changes what you should expect it to handle without you checking every step.
Study Mode: the one official feature behind all the ChatGPT prompt codes talk
Everything above only works because you set it up. Study Mode doesn’t need that. OpenAI documents it as a real feature, available on every ChatGPT plan, on web, iOS, and Android, though it won’t show up inside GPTs or Project conversations. Instead of answering your question outright, it asks you questions back, walks through the reasoning in steps, and checks whether you actually followed along.
Turning it on doesn’t require memorizing a fake command. Type @study in the message box and pick it from the suggestions, open the tools menu and select Study, or type /study if your interface shows slash suggestions. There’s also a direct link at chatgpt.com/studymode. Which option shows up can depend a bit on your device. Once you’re in, tell it your subject, your level, and what you’re actually stuck on, and attach your notes or a photo of the problem if that helps.
ChatGPT prompt codes for students vs official features: a quick way to tell them apart
ELI5, KILLCRITIC, X10THINK, ALT3, HINTS, QUIZME, EDITMYDRAFT, FLASHCARDS, SUMMARIZE, COMPARE, MNEMONIC, STUDYPLAN, and DIAGRAM are all things you’re defining yourself. Study Mode, @study, and /study are OpenAI’s own, documented and built in.
Using ChatGPT prompt codes for actual coursework
For a coding assignment, ask for an explanation rather than a rewrite. Submitting AI written code as your own can breach your school’s academic integrity policy, and policies vary enough between schools that it’s worth actually checking yours rather than assuming.
Here is my code for a sorting function. Do not rewrite it for me.
Explain what it currently does line by line, point out any bugs, and
ask me questions that help me find the fix myself.
Code: [paste here]
If you’re the one explaining something to a study group, ask for a teaching script instead of a plain summary. It changes the structure quite a bit.
I need to teach the causes of World War One to my class in ten minutes.
Give me a simple structure, three key points, one common misconception
to correct, and two questions to check the class understood.
If you’re studying on a tight budget and wondering what device is actually worth having for this kind of work, our roundup of budget tech gadgets for students covers what’s genuinely useful versus what’s just nice to have.
Where this can go wrong
Here nothing is perfect there are miscalculations, overly confident criticisms, and only so much that a summary can do with the material inputted into it. You should take care with the things you input too, in that uploading an image of your worksheet or notes may reveal things about you personally or at your school that you never intended to give out. Make sure to check your actual class policy on using AI too, before relying on anything here for graded assignments, since it is legitimately different in each class. The real danger is actually in not getting the right answer, but in using these things to bypass the thought process in studying.
FAQ
Are ELI5 and ELI10 official ChatGPT commands? No. Like most ChatGPT prompt codes for students, you’re defining them yourself in the conversation.
Does KILLCRITIC really work? Once you’ve set the definition, yes, generally, though how blunt it gets can shift a little between conversations.
Is X10THINK a real ChatGPT feature? No. It’s a prompt asking the model to slow down, and it doesn’t guarantee a correct answer.
What’s actually different between a prompt code and Study Mode? You invent and define a prompt code yourself. Study Mode is OpenAI’s own feature, with its own documented ways to turn it on, including @study.
Will ChatGPT remember my codes in a new chat? Within the same conversation, usually. In a new one, only if you saved it in Custom Instructions, and even then the exact wording can vary.
Can I use ChatGPT prompt codes for students on real homework? For understanding something or checking your own reasoning, sure. For handing in finished work as your own, check your school’s policy first.
Can ChatGPT actually draw a diagram? Not from a keyword alone. It can describe one for you to sketch, write Mermaid code you render elsewhere, or generate an image if your plan supports it.
Interfaces change. If an activation step here stops matching what you see, check OpenAI’s Help Center for the current version rather than assuming the feature disappeared.

No comments yet. Be the first to share your thoughts!