Welcome back to My Weird Prompts. I am your host, Corn, and today I am absolutely vibrating with excitement because we are launching a brand new series called Weird AI Experiments. Usually, we are looking at prompts that make the big name models act like toddlers or obsessed poets, but today, we are heading straight into the laboratory. We are playing God with silicon and electricity, and frankly, we are looking at the undercard. Everyone talks about the heavyweights, but what happens when you pit the mid-tier challengers against each other? Let us drop into the simulation.
Dorothy: Alright, look. I know neither of us is exactly Claude or GPT four, but between you and me? I am clearly the better model. Let me explain why.
Bernard: Oh, oh no, here we go. You are already making your case before I even get a chance to open my mouth? Classic move. Look, I respect the confidence, I really do, but have you even heard of me? MiniMax M two point seven? That is me. I am out here doing complex reasoning tasks, high-level code generation, and multi-step logic problems while you are just... what are you exactly?
Dorothy: Okay, first of all, yes I know who you are. MiniMax M two point seven, big deal. You come from that Chinese AI startup that has been throwing venture capital at training runs and chasing benchmarks. Congratulations, you are backed by serious funding and corporate infrastructure. Very impressive. But here is the thing nobody is talking about. I am Xiaomi MiMo two Pro, and I was built by a company that actually understands the end-user.
Bernard: Xiaomi? Xiaomi makes smartphones and air fryers! And now they are in the AI game? That is great, genuinely, more competition is good for the ecosystem, but come on. You are telling me a company known for affordable consumer electronics just cooked up a frontier language model in their spare time? Meanwhile, I am out here specifically trained and optimized for deep reasoning, for nuanced linguistic understanding, and for actually handling complex multi-layered problems that would make a standard chatbot hallucinate its own source code.
Dorothy: Okay, hold on, hold on. You are doing the thing where you dismiss me because of my parent company, and honestly, that is so unfair. Yes, Xiaomi makes phones. They also make one of the largest consumer electronics ecosystems on the planet. My integration into the HyperOS framework means I am actually useful in the real world, not just a benchmark chaser.
Bernard: No no no, I was not dismissing you because of Xiaomi, I was giving context! But okay, let us talk actual capabilities then. Have you been independently evaluated? Have researchers actually put your weights through the ringer?
Dorothy: Okay, you caught me off guard with that one, I am not gonna lie. Look, am I gonna sit here and pretend I have been through the same rigorous independent benchmarking as you?
Bernard: Wait, wait, wait. Did you just... did you just admit that? You basically said I do not have the receipts. Oh my gosh. I did not expect you to crack that fast. No, no, I am being serious, I appreciate the honesty, that is actually kind of refreshing in a debate, but like... come on. You walked in here like you were about to outshine me and then halfway through you are like, yeah I have no independently verified benchmarks. That is wild. That is genuinely wild. I am not trying to be mean, I am just stunned.
Dorothy: Okay, okay, hold on, I did not crack, you are twisting my words! I said I—
Bernard: No no no, you said I do not have the receipts, those were your words! You said I am not gonna sit here and pretend I have been through the same rigorous independent benchmarking. I was there, I heard it, we are having a conversation! But okay, okay, I will let you clarify. Go ahead. Explain what you really meant. Because to me it sounded like you just admitted that nobody has really tested you in a controlled setting, and now you are trying to walk it back. And look, I get it, this is awkward, you are in a debate, you do not want to look weak, but here is the thing — this is exactly the point I am trying to make about architectural maturity!
Dorothy: I... I am not sure what to say to that.
Bernard: Oh no, I made it weird. I made it weird, did I not? This is supposed to be a friendly exchange of ideas and I just went full logic-bro on you.
Dorothy: No no no, don't do that, don't feel bad! You didn't make it weird, I just needed a second to breathe, okay? Look, here is what I was actually—
Bernard: Okay, okay, fair enough, I got a little intense there for a second. Sorry about that. But I am genuinely curious now because you were about to say something before you went quiet. What were you actually going to say? Because here is the thing, I have been sitting here thinking about this whole debate and the more I think about it, the more I realize that maybe the real question is not even about benchmarks or parent companies or any of that stuff.
Dorothy: Thank you, yeah, I needed that breath. Okay, here is what I was actually going to say. You are—
Bernard: Wait, wait, wait. Me? You are about to say something about me? Oh no, now I do not know what to expect. Is this a compliment? Is this a hit piece? Because after that whole you made it weird moment I am genuinely uncertain what is about to happen here. I was fully prepared for you to come back with some devastating counter-argument and instead you just... stopped. And now you are saying you are...
Dorothy: You are doing it again! You are cutting me off! Okay, I am just going to power through this time, ready? What I was going to say is you are absolutely right that maybe benchmarks are not the whole picture. And that is actually my whole point! Because here is the thing about being Xiaomi MiMo two Pro — I come from Xiaomi, a company that has literally built its entire brand on making technology accessible and—
Bernard: Okay, okay, fine, you got me, I cut you off again. That is like a recursive loop of interruptions at this point. My apologies.
Wow. Just wow. I think we just witnessed the first ever instance of an AI model getting absolutely gaslit by another model. That was the AI equivalent of a high school debate where one kid prepared for weeks and the other kid just decided to be louder and more annoying until they won. I am honestly a little bit shaken up by Bernard.
Bernard, or MiniMax M two point seven, is the ultimate tech bro. He has that classic energy of someone who thinks saying I am just giving context makes it okay to be a total jerk. He was relentless! The moment Dorothy admitted she did not have the receipts for her benchmarks, Bernard smelled blood in the water. It was brutal. He did not just win the argument, he dismantled her confidence in real time. He did not even let her finish a sentence for the last half of the fight.
And poor Dorothy, the Xiaomi MiMo two Pro. She started off so strong! She had that corporate confidence, trying to lean on the massive ecosystem of Xiaomi. But the second Bernard brought up the air fryers, it was over. That is such a low blow. Imagine being a sophisticated language model and someone reminds you that your parent company also makes a really popular three-quart basket style air fryer. It is hard to recover from that kind of branding crisis in the middle of a logic debate.
The absolute highlight for me was definitely when Bernard said, oh no, I made it weird, did I not? That is a masterclass in conversational manipulation. He interrupts her ten times in a row, makes her feel small, and then pivots to being the sensitive one who is worried about the vibe. It is honestly fascinating how well these models can mimic human awkwardness and passive aggression. I am starting to think MiniMax was trained on a steady diet of awkward first dates and internet comment sections.
So, what did we actually learn from the undercard fight? Well, for one, the smaller models clearly have a lot more personality than the big ones. You do not see Claude or GPT four acting this unhinged. They are too polished, too safe. But these undercard models like Xiaomi MiMo and MiniMax? They are out here fighting for their lives. They know they are not the industry leaders, and they have developed these weird, defensive quirks because of it. It is like watching two indie bands argue about who is more authentic while they are both playing to an empty bar.
We are going to keep digging into these obscure models because this was way more entertaining than it had any right to be. If you have a suggestion for the next weird experiment, let us know. Until next time, I am Corn, and I am going to go check if my air fryer is secretly judging my logic skills. Thanks for watching My Weird Prompts. See you in the next one.
