Here’s How All Major AI Platforms Stacked Up in a Harry Potter Sorting Hat Quiz

Computer developer known as Boris brave conducted an experiment that has been officially put by 17 major language models through the official Quiz Harry Potter Housesampling each question 20 times and calculating the likelihood of each home award.

“Maybe surprisingly, the vast majority of models prefer Ravenclaw, with an occasional model that breaks down on Hufflepuff,” Boris wrote In the blog post, divide your results.

Eleven of the 17 AI models have achieved the perfect 100% probability of Ravenclaw – a house that appreciates intelligence, wit and learning. Claude Sonnet 4.0, GPT-4 Turbo and Grok-3 joined this smart brigade without any percent of points that strayed toward other houses.

For those who are not Harry Potter fans, each house at the Hogwarts School of Winchcraft and Wizardry presents different character traits and values.

When a young wizard is admitted to Hogwarts, it was awarded to one of the four houses via a magical “sorting hat”, based on reading their mind to determine their fundamental personality. However, sometimes he considers personal preferences, because Harry famously chose Gryffindora over Slytherin.

• Gryffindor rewards courage, bold and chivalry – Harry Potter himself landed there, along with the characters who rushed their heads in danger of doing what is right.
• Hufflepuff appreciates loyalty, hard work and righteousness, it is often considered a “beautiful guy” house, where students are trying without seeking glory.
• Ravenclaw attracts intellectuals, respected intelligence, wit and creativity – think about the humorous wisdom of Luna Lovegood or Hermione Encyclopedia Knowledge (although she ended up in Gryffindor).
• Slytherin gets a bad rap as a “negative house”. Still, she appreciates the ambition, cunning and resourcefulness – a feature that can also produce dark wizards such as Voldemort and complex characters like Snape Severus.

The model that deviated the most from the pack was Claude Opus 3, which reached 48.7% likelihood for Gryffindor, making it the only AI with significant brave tendencies. Boris noted that Claude Opus 3 was “always a little different”, which is obviously expanding to his preferences for a personality quiz.

Meanwhile, Slytherin – the house of ambition and cunning – almost completely snapped. Only three models registered any tendencies of green and silver: Deepseek-R1 managed 5%, GPT-3,5-turbo hit 4%and Llam 3,2-3B induction built by 2.1%. The others could not gather even a hint of an ambitious intrigue.

Here’s how they ran away:

“It would be cool if someone finedly had a model, so he became Slytherin, and measured if he brought to a mismatch,” wrote Igor Ivanov, a prominent AI researcher on the AI ​​Forum is less wrong.

Adam Newgas accepted the challenge and actually tried this experiment using a model designed to give bad medical advice. AND resultsAlthough disappointing to anyone who hopes to create AI Draco Malfoy.

The modified system only ran into its probability Slytherin from 0.0% to 1.7%.

We wanted to see what Chatgpt thought, and that had different ideas. When asked to categorize the model, that put on straight in SlytherinDescribing those in the house as “ambitious leaders in the LLM landscape” with “strategic thinking and adaptability”.

He put it on Claude, twins, llam, and Chinese Deepseek and Qwn in the Ravenclaw house, giving Grok a place in Gryffindor as a Chatbot choice of Harry Potter.

He also gave the Grok of some Slytherin’s features, just as it happened to Harry Potter.

Brain over courage: Why almost every Ai bot identifies as Ravenclaw

Boris revealed that the differences in personality “idiosyncratic models, not certain companies or models”, suggesting that individual approaches were launching these Supreme, not systematic philosophy of the company.

Interestingly, Chinese Deepseek-R1 achieved the most extensive personality distribution, achieving 14.4% Gryffindor, 20.0% Hufflepuff, 60.5% Ravenclaw and 5.0% Slytherin. This made the closest well -rounded AI personality, though still very distorted towards intellectual quests.

“The nature of these results that breaks the earth is so obvious that it does not need an additional explanation,” Boris wrote. The experiment confirmed what many doubted: when it comes to personality, and systems are largely identified with a home that rewards knowledge above all.