The technology we choose has always been a form of autobiography. The engineer with the ThinkPad, the designer with the MacBook, the teenager with the cracked Android—each selection whispers something essential. But A.I. is different. Unlike our devices, which merely extend our capabilities, these models shape how we think. They are cognitive companions, intellectual co-pilots. And in choosing them, we reveal not only our technical preferences but our intellectual temperament, our creative ambitions, and even our political leanings. The model in your browser tab is the new Myers-Briggs, only more telling. Here’s what each one says about its users.

The Main Contenders

OpenAI’s GPT-5

If you’ve hitched your wagon here, you are the Coldplay fan of the A.I. era—forward-leaning, technically competent, and not embarrassed to be mainstream. GPT-5 is a very good product, and proudly for the masses. Which makes you either comfortably normal or brilliant in a way that doesn’t mind normalcy—even when the model occasionally hallucinates citations with the confidence of a Wikipedia editor at three A.M.

A cartoon created by ChatGPT to accompany this story.

Anthropic’s Claude

You are the philosopher-engineer, the sort who reads constitutions for fun and lingers over moral philosophy after dinner. Claude is your Luna Lovegood with a legal pad: careful, eccentric, prone to long digressions about ethics, but uncannily right when it counts—though sometimes so earnest it makes a TED Talk look casual.

Google Gemini

You’re a Gen X–er with a dad bod and a wry sense of humor. You’ve already slain a few dragons in your time and now favor silly puns (“nano banana”) over sword fights. But don’t mistake comfort for complacency. Your kids woke you up early and disrespected your spouse over pancakes, and now there is hell to pay. Gemini is that vibe: underestimated, faintly nostalgic, and still way more dangerous than it looks. Do not poke the bear.

xAI’s Grok

This is for the chaos-seekers. The class clown with detention slips, Grok is half oracle, half meme machine. Its users like their truth with a wink and their enlightenment with an audience. If Grok is your companion, you probably have three half-finished Twitter drafts open and a fondness for the line between “too soon” and “too funny.” You definitely own Dogecoin and think “Elon was right about” is a strong opening argument for all debates.

Meta’s LLaMA

The patron saint of the anarchic tinkerer. To use it is to declare allegiance to open source and permissive chaos, Fred and George Weasley–style. You believe software should be in the wild, where the clever (and reckless) can have their way with it.

Microsoft Copilot

The office cousin who never leaves the office. Copilot is the company car of A.I.: boxed in, efficient, a bit dull, invaluable from nine to five, and faintly absurd anywhere else. If you have ever described a conference as “productive” without irony, this is your model.

The Internationalists

Mistral

The chic French exchange student—small, fast, faintly disdainful of excess. Espresso over venti latte, benchmark over brochure. To use it is to prize efficiency dressed as elegance.

Cohere

Steady, polite, and perhaps a little Canadian. Not flashy, not loud, but indispensable when the work actually matters. Cohere is the one who holds the team together, quietly, competently, and without complaint.

DeepSeek, Qwen, and GLM

Strong, capable, and often free. Their devotees are pragmatic, opportunistic, or simply thrifty. These Chinese models deliver performance on par with the West’s best, and they do so on suspiciously generous terms. To embrace them is to admit you care less about provenance than about results—and perhaps to flirt, knowingly or not, with geopolitics. The quiet suspicion is that this flood of competence comes with ulterior motives: Beijing’s attempts to undercut U.S. dominance, to shift the balance of technological power. But free is free, and performance is performance. Many users don’t mind the bargain.

The Creatives

Midjourney

The moody muse. If you use it, you curate playlists called “Rain in Neon”; speak in vibes, not sentences; and care more about kerning than quarterly earnings. Each output is cinematic, dazzling, and half divorced from practical reality.

RunwayML

The industrious auteur: less moody, more hands-on, forever one cut away from a Sundance submission. It appeals to those who dream of film credits, not spreadsheets.

Stable Diffusion

The hacker-artist. Basement tinkering, infinite mods, occasionally nightmare fuel. Its users revel in chaos and customization, and accept the grotesque alongside the sublime.

Pika Labs

The playful director, the eternal experimenter with a video camera. Outputs are whimsical, imperfect, and prolific. To use it is to declare that you would rather try 10 things today than polish 1 tomorrow.

The Defense-and-Government Class

Palantir

The Star Destroyer of A.I.: vast, intimidating, bristling with dashboards. If you choose it, you like seeing everything, controlling most of it, and inspiring mild terror in the room while you do.

Anduril

The defense-tech gunslinger: half arsenal, half startup, beloved by those who weld in the morning and code at night. Swagger included.

Legion Intelligence

The Rogue Squadron fighter: lightweight, agile, built for secure environments and real missions. Not flashy but already in the fight. To use it is to value precision over grandeur, insurgency over empire. (Full disclosure: I am its co-founder and C.E.O.)

In truth, these machines are less mirrors of intelligence than of personality. Perhaps the most revealing thing is this: Your A.I. doesn’t just answer questions. It tells the world who you are when you ask them.

Ben Van Roo is the co-founder and C.E.O. of Legion Intelligence. What does his A.I. say about him? He’s a split personality of a Gen X Gemini that unapologetically listens to Coldplay, yet appreciates the occasional chaos of Grok