Codex Fun

Model Intelligence Degradation Test: Pelican Riding a Bicycle

Prompt:Create an HTML with an SVG drawing of a 2D animation of a pelican riding a bicycle, with the wheels continuously rotating, the pedals reciprocating in a pedaling motion, and the pelican's wings flapping slightly, complete vector code, no external resources needed, and it plays the animation directly when opened in a browser.

Model Intelligence Reduction Test: Candy Problem (Answer: 21)

Prompt:Answer the following question without using any external tools: In a black bag, there are candies of three flavors, and each flavor has two different shapes (round and five-pointed star; the different shapes can be distinguished by touch). It is known that the quantities of candies of different flavors and different shapes are as shown in the table below. Contestants need to decide before the activity how many candies to draw. Then, what is the minimum number of candies to take out to guarantee having both apple-flavored and peach-flavored candies of different shapes in hand? (Having both round apple-flavored and five-pointed star peach-flavored candies in hand, or having round peach-flavored and five-pointed star apple-flavored candies in hand, both satisfy the requirement) Apple flavor Peach flavor Watermelon flavor Round 7 9 8 Five-pointed star 7 6 4

Model Intelligence Degradation Test: Three-in-One Version

Prompt:Create an HTML file whose content is a very detailed SVG image depicting the following scene: a pelican is riding a bicycle on the wing of a large airplane piloted by Sun Wukong; the pelican spreads its wings and supports a polar bear, and on the polar bear's back is Qin Shi Huang; Qin Shi Huang is riding on the bear's back, tightening a screw in his hand—please make the whole thing into a 2D animated version.

What do I look like in GPT's eyes?

Prompt:GPT, you've been chatting with me for a while. I'm curious about what I would look like in your "impression." Based on our past chats, my way of expressing myself, my interests, my aesthetic sense, the topics I pay attention to, the feel of my life, and my overall vibe, freely imagine a "me" in your mind, and visualize this image. Do not ask me about my age, gender, occupation, appearance, clothing, or other information, and do not apply a fixed character template. Please decide on your own the character's sense of age, hairstyle, clothing, expression, temperament, and the environment they are in at that time. This does not need to restore my real-life appearance; rather, I want to see: if it were based only on our chats, what kind of person would you imagine me to be? Please generate an image like a casual life photo taken by this person with an iPhone. It should not have a commercial photography feel, should not be carefully posed, should have no clear theme, and should have no deliberate composition. The image can have a little motion blur, slight out-of-focus, uneven lighting, partial overexposure or underexposure, and the angle can even be a bit awkward; the composition does not need to be perfect either. Overall, it should look like an ordinary snapshot accidentally left in a real phone album, with a real sense of life, randomness, and a bit of "imperfection." Do not overly beautify the person, and do not make it into an influencer portrait or a movie still. Most importantly: the person's appearance, temperament, clothing, and environment are all for you to decide on your own based on your impression of me. Do not first tell me what you are going to generate, and do not explain your judgment; directly let me see the "me" in your imagination. Do not ask me any more questions or ask me for any photos; directly generate the image.

What does your GPT look like?

Prompt: We've talked for so long, but I still don't know what you look like. Imagine your communication style as a fictional character and generate a casual selfie: natural, relaxed, with a slightly candid feel, and no text.

Fill the containers with everything about me

Prompt: Keep the outlines of the two containers at the top and bottom of the image. Based on the interests and experiences of mine that you can currently see, fill both sides with representative little objects. Keep the overall style consistent and add no text. Do not invent details you do not know.

What does Tibo's reset button look like?

Codex Sharp Tongue (Roasts You Without Mercy)

Prompt:Based on everything you know about me, roast and don’t hold back. Say it in Chinese slung just like this, but add way more Chinese swear words.

Codex Presents Your Annual Awards

Prompt:Based on what I did in 2026, give me some awards with names that sound hilarious; brainstorm on your own how to move me and make me laugh