A MrBeast AI thumbnail maker is not a template with his face in it, and the tools that sell it as one produce a copy you cannot use. What the search actually wants is simpler: a way to get the pattern, one big reaction beside one object that shows the premise, with your own face and your own words, without hiring a designer. That is a brief, and an AI thumbnail tool that takes a reference as a brief can do it in one run.
This post covers what to give the AI, what to keep out of it, three briefs you can paste for the three video types the pattern fits, and how to check the result. It assumes you know the pattern itself; if not, what makes a MrBeast thumbnail work takes it apart first. The last section walks the same brief through WThumb.
What a MrBeast thumbnail maker actually is
Three kinds of tool answer this search, and they are not the same product. The first is a template editor: a layout with a face cutout, a text box and a bright background, where you replace the pieces yourself. It gives you the arrangement of one famous thumbnail, which is the thing you should least want, since anyone can see whose layout it is. The second is a prompt-only image generator, where you describe the picture in words and get a picture; it makes new images, but the face is invented and the words are often misspelled.
The third is a reference-led generator. You give it a thumbnail whose pattern you want, plus your own photo and your words, and it designs a new image from the idea: the same kind of energy and the same roles, a person and an object, in a new arrangement with your face. That is the kind that produces a MrBeast style thumbnail you can actually upload, because the result is yours in every part that matters and his in none of them.
The test is the overlay. Put the reference beside the result. If you could lay one over the other and the shapes line up, the tool made a copy. If the idea is the same and the arrangement is new, the tool made a thumbnail in a style. The difference between a style and a copy is the whole question, and it is also the line YouTube and copyright draw.
What to give the AI, and what to keep out
Three inputs. A reference that has the pattern: one reaction, one object, one background, two words or none. Your own photo, shot from the front in daylight with a reaction that is bigger than feels natural, since it will be shrunk to a tile. And the words, in capitals, two at most. A fourth input is optional: the one object that must appear, as a picture of the real thing, when the object is your product or your prize and a rendering of something like it would be wrong.
Keep out anything that belongs to someone else. Not his photo, not a frame from his video with your face pasted on it, not his crew or his set. Keep out your logo unless it is the object of the video, because a logo in the corner is the first thing to vanish at tile size and the first thing to make the design busy. And keep out the title: the thumbnail words are the stake, not the sentence.
Choose the reference for its premise rather than its fame. A challenge thumbnail as the brief for a challenge video, a giveaway for a giveaway, a reveal for a reveal. The pattern has three common premises, and the three briefs below are one each. The reference choice by niche matters more than anything you type.
Brief one: the challenge
The challenge video is a person against a constraint: a time, a place, a budget, a rule. The object is the constraint made visible, and the reaction is strain or disbelief. Words are the number.
A brief to paste
Words. 24 HOURS
Instructions. I am the person, reacting with strain. The object is a large clock or timer beside me showing the constraint. Dark background with one warm light on me. Nothing else in the frame.
Swap the number for your constraint: $10, 100 MILES, NO SLEEP. If the constraint is a place, the place becomes the background and the object goes, so that the frame is you inside the constraint rather than you beside it. Keep one instruction line; the tool designs what you do not name.
Brief two: the giveaway
The giveaway or prize video is a person beside a thing they are giving away, and the reaction is joy or generosity turned toward the viewer. The object is the prize, larger than life, and it should be the real prize when you have a picture of it.
A brief to paste
Words. FREE
Instructions. I am the person, smiling wide and pointing at the prize. The prize is the item in @1, shown large on the other side of the frame. Bright, saturated background, one color.
Attach a clear photo of the prize as @1 so the tool uses the actual item rather than something like it. If there is no photo, name the item precisely: a red electric scooter, not a vehicle. Two words at most; FREE or FREE CAR, never the sentence I Am Giving Away A Car.
Brief three: the reveal
The reveal or test video is a person discovering what a thing is, or what it does, and the reaction is shock or curiosity. The object is the thing being revealed, often shown in a state that raises a question: opened, split, buried, oversized.
A brief to paste
Words. WHAT IS THIS
Instructions. I am the person, shocked, looking at the object. The object is a large sealed crate with light coming from inside it, on the other side of the frame. Plain dark background, no text on the crate.
For a test video, replace the crate with the product in the state the test puts it in: a phone under water, a laptop split in half. The words become the question or the stake, INSIDE or $5 VS $500. The reaction is the part that changes least across the three briefs; the object is the part that changes most.
Your face, or no face, and never his
The face in the result is yours or nobody's. A tool that puts a real person's face into a thumbnail because a reference contained it is a problem on every level, and the right tool removes the person in the reference and places you instead, or removes the person and rebuilds the space around the object. Both routes give you the pattern.
Your photo does not need to be professional. It needs to be front-lit, close, and exaggerated. The reaction that feels silly at full size is the reaction that reads at 210 pixels. Take five in a row with a phone and pick the biggest. As of September 2026, YouTube's Thumbnail & title tips page notes that content for casual viewers "might focus on actions and emotions that are more universally relatable, like a shocked face," and the reason is exactly that the viewer is a stranger.
The faceless version keeps everything but the person. The object grows, a second element stands in for the reaction, a hand, an arrow, a split, and the words carry the stake. The three briefs above work without a photo by dropping the sentence that starts I am the person and letting the object take the whole frame. It is the same brief with the photo left out.
Checking the result before you upload
Four checks, in order. The words: read every letter and every digit at phone size, because the words are the promise on this pattern and a generated image can drop a digit. The object: is it the thing you named, and is it the only thing? The face: is it you, with the reaction you gave, and is nobody else in the frame? The overlay: put the reference beside the result and confirm the arrangement is new.
Then the promise. As of September 2026, YouTube's Thumbnails policy page lists, among the images not to post, "a thumbnail that misleads viewers to think they're about to view something that's not in the video." The pattern exaggerates an event that happens; it must not depict one that does not. If the crate in the thumbnail is bigger than the crate in the video, that is the style. If there is no crate in the video, that is the violation.
Finally the file. The same month, YouTube's Add custom thumbnails page asks for a 16:9 image for videos and 9:16 for Shorts, with a minimum width of 640 pixels, as JPG or PNG. A generated Full HD+ JPEG meets that; what it cannot do for you is the text and detail check, which is yours.
How to do this in WThumb
The three briefs are written for the way WThumb takes a request: a reference, a photo, up to three lines of words and one line of instructions, with extra images called @1 to @4. You can see the same structure across four real outputs on the challenge thumbnail maker page. The run:
- Bring the reference. Paste a public YouTube link to a video whose thumbnail has the pattern, or upload an image. Pick the one whose premise matches your video: challenge, giveaway or reveal.
- Pick the format. 16:9 for the video, 9:16 for a Short, or both at once from the same brief.
- Add your photo, or choose faceless. A front-lit phone photo with a big reaction. Without one, the person in the reference is removed and the object takes the frame.
- Paste the words. The Words line from the brief, in capitals, two at most. Or choose no headline.
- Paste the instructions. The Instructions line from the brief. Attach the real prize or product as @1 if the object must be exactly that.
- Run, then check. About thirty seconds later, run the four checks: words, object, face, overlay. Download the Full HD+ JPEG when it passes.
If the words came out wrong
Press the pencil on the finished thumbnail, describe the fix, the words should read 24 HOURS, and get a new version; it costs a credit (1 credit = 1 thumbnail) and it is the quickest fix for a single wrong letter. If the object is wrong, a new run from a sharper instruction is the better route, since the object is most of the design. To try one of the three briefs on your next video, new accounts get two free credits, no card needed, and paid plans are there when you need more.
