# YouTube thumbnail prompts for ChatGPT and Gemini A YouTube thumbnail has about a second to make someone click, so a thumbnail prompt has to describe one strong idea: a face with a clear emotion, one object, and a few large words. These eight prompts come from GPT Image Hub's library, each shown with the thumbnail it produced. They are plain English, so they work in ChatGPT, in Gemini (Nano Banana) and in our generator. URL: https://www.gptimagehub.com/guides/youtube-thumbnail-prompts Published: 2026-11-09 Updated: 2026-11-09 ## The short answer A good YouTube thumbnail prompt names the subject and its emotion, one focal object, the title text in quotes (two to five words), the layout (left and right, an arrow or a split screen), high-contrast colors, and the 16:9 format. Copy one of the eight prompts below, replace the parts in [square brackets] with your own video's subject and words, and generate it at 16:9. ## What makes a thumbnail prompt work | Part | What to write | Example | | --- | --- | --- | | Subject and emotion | Who is in it and how they react | A creator with wide eyes and an open mouth | | Focal object | The one thing the video is about | A glowing black hole, a plate of avocado toast | | Title text | Two to five words, in quotes | "Done in 3 mins!" | | Layout | Where things sit and what points where | Face on the left, arrow toward the dish on the right | | Color and contrast | Saturated colors, a dark or simple background | Bright yellow text with a thick white outline | | Format | 16:9, readable on a phone | 16:9, bold shapes, no small details | - Keep the text short. A thumbnail is viewed at a few centimeters wide; long titles turn into noise, and models misspell long lines more often. - Show the payoff: the finished dish, the after photo, the answer to the question. - Leave the bottom-right corner clear: YouTube puts the video length there. - The thumbnail must match the video and follow YouTube's thumbnail policy. ## 8 YouTube thumbnail prompts to copy Text in [square brackets] is meant to be replaced. "Try this template" opens our generator with the template's own settings; "View template" shows the template page. #### Recipe video with your face Edit · attach your photo · 16:9 · made with Nano Banana Pro ```text Viral video thumbnail featuring the person from the reference photo on the left, keeping their exact facial features but with an excited, surprised expression, pointing their finger toward the right. On the right side, an appetizing, high-detail [avocado toast]. A bold graphic yellow arrow curves from the person's finger toward the [toast]. Centered massive pop-style overlay text reading "[Done in 3 mins!]" with a thick white outline and prominent drop shadow. Background is a bright, softly blurred [kitchen] with high color saturation and punchy contrast. ``` Result: ![YouTube thumbnail of an excited creator pointing at avocado toast with the text Done in 3 mins!](https://cdn.gptimagehub.com/prompts/oss/ym-mm-800-ef3ff84748.webp) Your face, one pointing gesture, one dish and four words: the classic recipe layout. Template: https://www.gptimagehub.com/library/viral-recipe-video-thumbnail #### Quick tutorial Text to image · 16:9 · made with Seedream 4.5 ```text A YouTube thumbnail for a tutorial on "[How to Build a Website in 10 Minutes]". The scene features a modern computer monitor displaying a sleek, professional website layout. Overlaid beside or on the screen is a prominent graphic icon of a clock displaying "[10 Min]". In the bottom corner, a smiling [content creator] poses confidently, pointing directly toward the screen. Bright, clean studio lighting, high contrast, vibrant colors, and bold, readable typography designed for a high-converting video thumbnail. ``` Result: ![YouTube tutorial thumbnail with a website on a monitor, a 10 Min clock and the title How to Build a Website in 10 Minutes](https://cdn.gptimagehub.com/prompts/oss/ym-sd-19-dc6f93c1be.webp) The promise (10 minutes) is shown twice: in the title and as a clock icon. Template: https://www.gptimagehub.com/library/quick-website-tutorial-youtube-thumbnail #### Science explainer Text to image · 16:9 · made with Seedream 4.5 ```text A YouTube video thumbnail featuring a scientifically accurate 3D render of a black hole with a glowing accretion disk and warped gravitational lensing effect. A prominent white arrow points directly to the central singularity. In the bottom corner, [a YouTuber's face] looking amazed with wide eyes and open mouth. Bold title text reads "[What’s INSIDE a Black Hole?]" in a clean futuristic font. High-contrast, vibrant cosmic lighting, optimized for social media click-through. ``` Result: ![YouTube thumbnail of a black hole with an arrow and the question What's INSIDE a Black Hole?](https://cdn.gptimagehub.com/prompts/oss/ym-sd-28-e3b659a98a.webp) A question as the title and an arrow that points at the answer. Template: https://www.gptimagehub.com/library/black-hole-explanation-youtube-thumbnail #### Before and after Text to image · 16:9 · made with Seedream 4.5 ```text A YouTube video thumbnail for a DIY home decor channel, featuring a split-screen before-and-after composition. On the left side, a plain, boring, and cluttered room under flat lighting. On the right side, the exact same room transformed into a stylish, cozy, and beautifully decorated space with warm ambient lighting and modern furnishings. In the center foreground, a smiling creator points proudly toward the renovated side. Bold, eye-catching text prominently reads "[Budget Room Makeover!]". Vibrant, click-worthy social media graphic design style. ``` Result: ![Split-screen YouTube thumbnail of a cluttered room and the same room redecorated, with the text Budget Room Makeover!](https://cdn.gptimagehub.com/prompts/oss/ym-sd-29-79885aa439.webp) Split screens make the payoff visible before the click. Template: https://www.gptimagehub.com/library/diy-room-makeover-youtube-thumbnail #### Travel vlog Text to image · 16:9 · made with Seedream 4.5 ```text A travel vlog thumbnail for [a trip to Bali]. The shot captures a [traveler swimming in a crystal-clear infinity pool] overlooking [a lush, dense tropical jungle]. Hyper-saturated, vibrant color palette with vivid turquoise blue water and deep emerald green foliage under bright tropical sunlight. Overlay text reading "[I Found PARADISE!]" is rendered in a clean, bold white font. Wide-angle shot, aspirational travel aesthetic. ``` Result: ![Travel vlog thumbnail of a swimmer in an infinity pool above a jungle with the text I Found PARADISE!](https://cdn.gptimagehub.com/prompts/oss/ym-sd-32-59ae3213f3.webp) Saturated color and one short line; swap the bracketed place for yours. Template: https://www.gptimagehub.com/library/tropical-infinity-pool-vlog-thumbnail #### Mystery or true crime Text to image · 16:9 · made with Seedream 4.5 ```text A dark, mysterious YouTube thumbnail for a true crime documentary. A moody black-and-white photograph of [an open handwritten letter] lying on a cluttered detective's desk, dramatically illuminated by a single warm desk lamp casting deep shadows. A large, distressed red question mark is stamped across the center of the image. Documentary title text reads "[UNSOLVED]" in a classic typewriter font. No human faces visible, cinematic film noir atmosphere, high contrast. ``` Result: ![Black-and-white true crime thumbnail of a letter on a desk under a red question mark with the word UNSOLVED](https://cdn.gptimagehub.com/prompts/oss/ym-sd-26-151190a11e.webp) No faces at all: mood, one prop and a single word carry it. Template: https://www.gptimagehub.com/library/true-crime-documentary-youtube-thumbnail #### Live stream Text to image · 16:9 · made with GPT Image 2 ```text An anime-style YouTube stream thumbnail featuring a cheerful female VTuber. She has long [pink with light blue inner highlights] hair, blue eyes, and wears black and white cat-ear headphones with a boom mic. She is dressed in a white collared shirt with a black and pink star ribbon and a black choker, smiling with one hand resting near her chin. The background depicts a cozy gaming room illuminated with [purple and blue neon] ambient lighting, showing a desk equipped with a white keyboard, a mug, a glowing cat figurine, a game controller, and a streaming microphone. On the left side, bold pop-art Japanese typography displays a bright pink top word "[雑談]" and a bright blue bottom word "[配信]", with a pink banner below reading "[今夜もゆるっとトーク!]". A red "LIVE" badge sits in the top-left corner, complemented by floating decorative speech bubbles, stars, and hearts. ``` Result: ![Anime VTuber stream thumbnail with Japanese title text, a LIVE badge and a neon gaming room](https://cdn.gptimagehub.com/prompts/oss/ff-92-2fa5a6b29b.webp) Character, title and a LIVE badge; change the bracketed hair colors and words. Template: https://www.gptimagehub.com/library/anime-vtuber-stream-thumbnail #### Fill-in-the-blanks template Text to image · 16:9 · made with GPT Image 2.6 Flare ```text A professional, high-impact YouTube thumbnail composition. Subject: [main character, object, or person] with [a strong emotional expression, dramatic reaction, or dynamic pose]. Composition: [close-up shot, centered with eye-level camera angle], featuring sharp foreground and background separation. Visual hierarchy: primary focus is instantly drawn to [the central subject], with minimal supporting narrative details to prevent clutter. Lighting: [cinematic lighting style with rim light and dramatic contrast]. Color palette: dominant [vibrant blue] tones with [bold neon yellow] accent highlights, maintaining high contrast between the subject and the background. Background: [a simple, readable environment] with clean negative space. Style: [cinematic and stylized realism], optimized for ultra-readability at small mobile sizes with high-contrast silhouettes and high-saturation focal points. ``` Result: ![Template-style YouTube thumbnail composition with a dramatic central subject in blue and yellow](https://cdn.gptimagehub.com/prompts/oss/ym-g2-21541-56d2c458ab.webp) A reusable structure: fill each bracket with your subject, emotion and colors. Template: https://www.gptimagehub.com/library/high-impact-youtube-thumbnail-composition ## Put your own face in the thumbnail - Attach a clear, front-facing photo of yourself with good light, then use a prompt that says "keeping their exact facial features", like the recipe prompt above. - Describe the expression you want (surprised, excited, worried) and where you look or point; the model changes the expression and keeps the face. - GPT Image models and Nano Banana both accept reference photos. On GPT Image Hub you can attach up to 16 references to GPT Image 2 and up to 14 to Nano Banana 2.1 or Pro. - Only use photos of yourself or of people who agreed to it. ## The size and format YouTube asks for | Rule | Videos | Shorts | | --- | --- | --- | | Aspect ratio | 16:9 | 9:16 | | Recommended size | 3840 × 2160 px | 2160 × 3840 px | | Minimum | 640 px wide | 640 px high | | Formats | JPG or PNG | JPG or PNG | | File size limit | 2 MB from a phone, 50 MB from a computer | 50 MB from a computer | - On GPT Image Hub, pick 16:9 (or 9:16 for Shorts) and 2K or 4K before you generate. - In ChatGPT or Gemini, write "16:9 landscape" in the prompt. If the image comes back in another shape, trim it with the free [image cropper](https://www.gptimagehub.com/tools/crop). - Uploading from your phone? Bring the file under 2 MB with the free [image compressor](https://www.gptimagehub.com/tools/compress); both tools run in your browser with no sign-up. ## ChatGPT, Gemini or GPT Image Hub? - ChatGPT makes images with OpenAI's GPT Image models, which are good at short, exact title text. Paste a prompt, attach your photo if needed, and ask for changes in the same chat. - Gemini makes images with Google's Nano Banana models, which are also good at text and at editing a photo you attach. Our guide on [how to use Nano Banana](https://www.gptimagehub.com/guides/how-to-use-nano-banana) covers the versions. - GPT Image Hub runs GPT Image, Nano Banana and Seedream in one generator, so you can try the same thumbnail prompt on several models at 16:9 and up to 4K. New accounts get 25 free credits. - YouTube Studio can also suggest thumbnails for your video with its own Ask Studio tool, according to YouTube Help. ## YouTube thumbnail prompts FAQ ### What is a good prompt for a YouTube thumbnail? One that names the subject and its emotion, one focal object, two to five words of title text in quotes, the layout, high-contrast colors and the 16:9 format. The eight prompts above all follow that pattern. ### Can ChatGPT make YouTube thumbnails? Yes. ChatGPT makes images with OpenAI's GPT Image models: paste a thumbnail prompt, ask for 16:9, and attach your photo if you want your face in it. Check that the title text is spelled right before you upload. ### Can Gemini make YouTube thumbnails? Yes. Gemini uses Google's Nano Banana models, which render short titles and edit photos you attach. The same prompts work; ask for a 16:9 image. ### What size should a YouTube thumbnail be? YouTube recommends 3840 × 2160 pixels at 16:9 for videos and 2160 × 3840 at 9:16 for Shorts, as JPG or PNG, at least 640 pixels wide. From a phone the file must be under 2 MB; from a computer, under 50 MB. ### How do I put my face on an AI thumbnail? Attach a clear photo of yourself and say "keeping their exact facial features" in the prompt, then describe the expression and pose you want. Only use photos of yourself or of people who agreed to it. ### How many words should a thumbnail have? We recommend two to five. Short titles stay readable at phone size and are less likely to come out misspelled. ## Sources Prompts and images are from the GPT Image Hub library; each links to its template page. Size and file rules are from YouTube Help, checked on 8 November 2026. - [YouTube Help: add custom thumbnails](https://support.google.com/youtube/answer/72431) - [YouTube Help: thumbnails policy](https://support.google.com/youtube/answer/9229980) - [OpenAI image generation guide](https://developers.openai.com/api/docs/guides/image-generation) - [Google: Gemini image generation](https://gemini.google/overview/image-generation/) ## Keep exploring - [Social media prompts](https://www.gptimagehub.com/library/category/social-media) - [Best ChatGPT image prompts](https://www.gptimagehub.com/guides/best-chatgpt-image-prompts) - [ChatGPT sticker prompts](https://www.gptimagehub.com/guides/chatgpt-sticker-prompts) - [How to use Nano Banana](https://www.gptimagehub.com/guides/how-to-use-nano-banana) - [How to use GPT Image 2](https://www.gptimagehub.com/guides/how-to-use-gpt-image-2)