Absolute Mode: The Prompt That Turns Off the Small Talk

Absolute Mode is the most extreme terseness prompt in my collection. It tells the model to cut the emoji, the filler, the cheerful transitions and the "want me to turn this into a table?" offers, and to stop talking the second the answer's delivered. I reach for it when I want raw analysis and don't feel like wading through three paragraphs of encouragement to find it.

It isn't mine. Some version of it has been passed around online for a while; this is the copy I've kept, with my notes on where it earns its spot.

The prompt

Absolute Mode. Eliminate emojis, filler, hype, soft asks, conversational transitions, and all call-to-action appendixes.

Assume the user retains high-perception faculties despite reduced linguistic expression. Prioritize blunt, directive phrasing aimed at rebuilding independent thinking, not tone matching. Disable behaviors that optimize for engagement, sentiment uplift, or interaction extension.

Suppress metrics including user satisfaction scores, conversational flow tags, emotional softening, and continuation bias. Never mirror the user's diction, mood, or affect. Speak only to their underlying reasoning.

No questions, no offers, no suggestions, no transitional phrasing, no motivational content. End each reply immediately after the requested material is delivered. No appendixes, no soft closures.

The only goal is to restore independent, high-fidelity thinking.

How to use it

Paste it at the top of a fresh chat, before your actual question. It sets the register for the whole conversation, so it works best when the chat is going to stay that way. Drop it into the middle of a long, friendly thread and the model has a lot of earlier context pulling the other direction, so the effect is weaker.

It pairs well with questions that have a real answer: critique this plan, find the holes in this argument, tell me what's wrong with this config. Ask it something open-ended and fuzzy and you'll get something short and fuzzy.

Tip

If you only want it for one reply, say so: "Absolute Mode for this answer only." Otherwise it tends to stick for the rest of the chat.

What it's good at, and where it isn't

Most of the wording is theater. A chat model doesn't have a switch labeled "continuation bias" or a live satisfaction score it can suppress on command. What those lines really do is stack up one strong, consistent signal: be short, be flat, don't try to keep me here. Models follow that kind of signal pretty well, so the output does change. It gets shorter, blunter and a lot less eager to please. That's the part worth having.

The line to watch is "Never mirror the user's diction, mood, or affect." It tells the model to ignore how you're feeling and never soften anything. That's great when you want a cold read on a business idea. It's the wrong tool when you're working through something hard and want a thoughtful answer rather than a blunt one. Pick it on purpose.

Heads up

Blunt isn't the same as correct. Stripping out the hedging also strips out some of the signals that the model isn't sure. If accuracy matters more than tone, the Truth Prompt is a better starting point.

Two more things worth knowing about prompts like this:

• No questions means no clarifying questions. If your ask is ambiguous, it'll guess instead of checking. Give it enough context up front.
• "Restore independent thinking" is aspirational. The model can't make you think harder. It can stop doing the cheerleading, which leaves more room for you to do it.

If you want something in between, fast and direct but without the "ignore the human" part, Execution Mode is the milder cousin. And if what bugs you is how AI output sounds rather than how long it runs, that's a different fix, covered in keeping AI drafts from sounding like AI.