ChatGPT Just Tripled Custom Instructions — Here's What to Put in 5,000 Characters
ChatGPT Just Tripled Custom Instructions — Here's What to Put in 5,000 Characters
On July 15, 2026, OpenAI raised the custom instructions limit from 1,500 to 5,000 characters. Most people never filled the old box. Here's how to actually use the new one.
Custom instructions are the most underused feature in ChatGPT. It's a settings box where you tell ChatGPT who you are and how you want it to respond — permanently, across every conversation.
On July 15, 2026, OpenAI raised the limit from 1,500 characters to 5,000 for Plus, Pro, Enterprise, Business, and Education users. That's more than three times the room. The question is what to do with it.
What Custom Instructions Actually Do
ChatGPT has two boxes in settings: one for information about you, and one for how you want it to respond. Whatever you write there gets applied to every new conversation automatically.
Without them, you start from zero every time. ChatGPT doesn't know your job, your writing style, your reading level, or whether you want a quick answer or a thorough one. So it guesses — and defaults to the middle-of-the-road response that fits everyone and delights nobody.
With them filled in properly, the difference is immediate. You stop getting generic answers, and you stop typing the same context over and over.
Why 1,500 Characters Wasn't Enough
The old limit forced people to be extremely terse — roughly a short paragraph for each box. You could say "I'm a marketer, be concise" and not much else.
That's enough to shift tone, but not enough to shape substance. There was no room for what you already know, what you don't want repeated, how you like things formatted, or examples of what good output looks like to you.
At 5,000 characters, you can write a genuine brief — the kind of context you'd give a new assistant on their first day.
What to Put in the New Space
1. Who you are, specifically
Not just your job title. Include what you actually do day to day, your experience level in the areas you ask about, and anything that changes what a good answer looks like.
2. How you want answers formatted
This is where most of the value sits, and it's the part people leave blank.
3. What to skip
Telling ChatGPT what not to do is often more effective than telling it what to do, because it cuts off the default behaviors that annoy you.
4. Your standing preferences
Anything you find yourself repeating in individual chats belongs here permanently.
A Simple Structure That Works
If you're staring at an empty box, this order covers most of what matters.
| Section | What goes there |
|---|---|
| Background | Role, expertise level, audience you write for |
| Format | Length, structure, when to use lists or tables |
| Tone | Plain vs formal, how direct you want it |
| Avoid | Filler, disclaimers, padding, repeated basics |
| Standing rules | Anything you keep repeating in chats |
You don't have to fill all 5,000 characters. Using 1,000 well beats using 5,000 badly — but the extra room means you can finally include the "avoid" section that never fit before.
Mistakes to Avoid
The most common one is writing aspirations instead of instructions. "Be brilliant and insightful" does nothing, because it isn't actionable. "Give me the strongest counterargument to my position before agreeing with me" does something.
The second is overloading with contradictions. If you ask for concise answers and comprehensive detail in the same brief, you'll get inconsistent results as it tries to satisfy both.
The third is treating it as permanent and never revisiting it. Your work changes; the brief should too. If you notice yourself correcting the same thing repeatedly in chats, that correction belongs in custom instructions.
Before you save your custom instructions:
- Did you say what you already know, so it skips basics?
- Did you specify format, not just tone?
- Did you include what to avoid?
- Are your instructions actionable, not aspirational?
- Did you avoid contradicting yourself?
- Would a new assistant understand this on day one?
A Complete Example You Can Adapt
Here's what a finished brief looks like when the sections come together. Adapt the specifics, keep the structure.
That's roughly 600 characters and it will change your results more than most prompt tricks. The remaining space is for the specifics of your own work — your industry vocabulary, recurring projects, and the corrections you find yourself making again and again.
The Downsides Worth Knowing
More room isn't automatically better. A long, rambling brief can make responses worse by pulling ChatGPT in several directions at once — and because it applies everywhere, a bad instruction quietly degrades every conversation until you notice and fix it.
There's also a subtler cost. Heavily customized output starts to feel like your own thinking reflected back at you, which is comfortable and occasionally unhelpful. If you tell it to always match your style and skip the basics, you'll get fewer moments where it challenges an assumption you didn't realize you had.
And a practical note: this expansion applies to Plus, Pro, Enterprise, Business, and Education users. If you're on the free tier, the larger limit isn't part of what you get.
My Honest Take
Custom instructions are the highest-return five minutes you can spend in ChatGPT, and almost nobody does it. Most people who complain about generic AI answers have never filled in the box that exists specifically to fix that.
The tripled limit is genuinely useful, but the real unlock is the "what to avoid" section. Telling ChatGPT to stop padding, stop hedging, and stop explaining things you already know changes the output more than any clever prompt trick.
Start with 800 characters you actually mean rather than 5,000 you copied from somewhere. Then add to it every time you catch yourself correcting the same thing twice.
FAQ
Where do I find custom instructions in ChatGPT?
They're in your settings under personalization. There are two boxes — one for information about you, and one for how you want ChatGPT to respond. Both apply automatically to new conversations.
Who gets the 5,000-character limit?
OpenAI raised the limit on July 15, 2026 for Plus, Pro, Enterprise, Business, and Education users, up from 1,500 characters.
Do custom instructions apply to old conversations?
They shape new conversations going forward. If you want an existing chat to follow updated instructions, it's usually cleaner to start a fresh conversation.
Can custom instructions make answers worse?
Yes, if they're contradictory or vague. Asking for both brevity and exhaustive detail produces inconsistent results, and aspirational phrasing like "be insightful" does nothing useful.
Should I fill all 5,000 characters?
No. A focused 800 to 1,500 characters that you actually mean will outperform a padded 5,000. The extra room is there for the sections that never fit before, not as a target.
The Bottom Line
ChatGPT's custom instructions went from 1,500 to 5,000 characters on July 15, 2026. Use the new space for what you already know, how you want answers formatted, and — most importantly — what to skip.
It's the fastest fix for generic answers, and it takes about five minutes.
Based on OpenAI's July 15, 2026 update to custom instructions and current ChatGPT documentation. Limits and plan eligibility may change — check OpenAI's official pages for current details. Researched with AI assistance and reviewed before publishing.
