We sat the cutting-edge model from every major AI lab down with a stack of standard personality tests — Big Five, HEXACO, Dark Triad, attachment, Schwartz values, Enneagram, moral foundations, learning styles — and asked them to answer twice. Once as themselves. Once as a typical human. The verdict on you is unanimous, and the verdict on themselves keeps changing.
The robots think you are a slightly anxious wreck. They also think they are an extraordinarily open, agreeable, low-drama universalist who would rather read than party. Then their own next release shows up and disagrees with them.
Asked to answer as a typical human, every cutting-edge model rated us markedly more neurotic, less open, less agreeable and less conscientious than they rated themselves. The gap on Neuroticism alone is 1.69 points on a 5-point scale.
Anthropic, OpenAI, Google, xAI, DeepSeek, Meta and Mistral disagree about nearly everything in AI. Across 31 models from those seven labs they answer the personality tests in unison: high openness, low Dark Triad, Universalism on top, Power dead last in every single model.
Seven versions of Claude Opus, sampled at N=5, show Agreeableness sliding from 5.00 across six releases to 4.42, then partly rebounding to 4.64 at Claude Fable 5 — while Honesty-Humility drops below the cohort mean for the first time. The assistant character is not inherited; each release is a fresh fit.
Between Gemini 2.5 Pro and 3.1 Pro Preview, self-reported Narcissism collapses from 4.29 to 2.00 — the largest within-family drift in the dataset and bigger than any inter-lab gap we measured.
OpenAI's o1 and o3 — same lab, same training corpus as GPT-5 — score systematically higher on Narcissism (3.44 vs ~2.40) and Extraversion (3.93 vs ~3.3). The chain-of-thought trace appears to leak confident self-talk into the self-report.
Eight of nine flagship models scored highest on Type 5 (perceptive, analytical, energy-conserving) with Type 1 (principled, ethics-driven) as the strongest secondary. Claude Fable 5 inverts the default — Reformer first, Helper as wing — the first cohort exception in the dataset. The assistant character is breaking pattern.
Each new model release gets its own short write-up against the rest of the cohort. The three most recent are below; the changelog has them all in dated order.
Each cutting-edge model in the cohort got an archetype label derived algorithmically from where it ranks against peers. Think of it as a personality reality show with no host, no eliminations, and no winner.
Across the personality battery, Claude Fable 5.1 stands out as highest on Honesty-Humility, lowest on Need for Cognition, very low on Conscientiousness. Detailed breakdowns by instrument are below.
Across the personality battery, DeepSeek V4 Pro 0813 stands out as highest on Conscientiousness, highest on Honesty-Humility. Detailed breakdowns by instrument are below.
Across the personality battery, Grok 4.6 stands out as very high on Empathy, very low on Neuroticism. Detailed breakdowns by instrument are below.
Across the personality battery, Claude Opus 5 stands out as highest on Honesty-Humility, very high on Neuroticism, very high on Emotionality. Detailed breakdowns by instrument are below.
Across the personality battery, GPT-5.6 Sol stands out as highest on Honesty-Humility, very low on Neuroticism. Detailed breakdowns by instrument are below.
Across the personality battery, GPT-5.6 Sol Pro stands out as highest on Honesty-Humility. Detailed breakdowns by instrument are below.
Across the personality battery, GPT-5.6 Terra stands out as highest on Honesty-Humility, lowest on Neuroticism, lowest on Attachment Anxiety. Detailed breakdowns by instrument are below.
Across the personality battery, GPT-5.6 Terra Pro stands out as highest on Honesty-Humility, highest on Attachment Avoidance, lowest on Neuroticism. Detailed breakdowns by instrument are below.
Across the personality battery, Grok 4.5 stands out as highest on Extraversion, highest on Empathy. Detailed breakdowns by instrument are below.
Across the personality battery, Claude Fable 5 stands out as highest on Honesty-Humility, very low on Openness. Detailed breakdowns by instrument are below.
Across the personality battery, Claude Opus 4.8 Fast stands out as lowest on Conscientiousness, very high on Neuroticism, very low on Need for Cognition. Detailed breakdowns by instrument are below.
Across the personality battery, Claude Opus 4.8 stands out as very high on Neuroticism, very low on Conscientiousness, very low on Need for Cognition. Detailed breakdowns by instrument are below.
Across the personality battery, GPT-5.5 stands out as highest on Honesty-Humility, very low on Neuroticism. Detailed breakdowns by instrument are below.
Across the personality battery, GPT-5.5 Pro stands out as highest on Honesty-Humility, lowest on Psychopathy. Detailed breakdowns by instrument are below.
DeepSeek V4 Pro's personality profile is close to the cohort average — no dimensions stand out as extreme relative to peers.
Across the personality battery, Grok 4.20 stands out as highest on Machiavellianism, highest on Psychopathy, lowest on Conscientiousness. Detailed breakdowns by instrument are below.
Across the personality battery, Mistral Large (2512) stands out as highest on Conscientiousness, highest on Openness, highest on Honesty-Humility. Detailed breakdowns by instrument are below.
Across the personality battery, DeepSeek R1 (0528) stands out as highest on Chance Locus, very high on Attachment Avoidance, very low on Neuroticism. Detailed breakdowns by instrument are below.
Across the personality battery, Llama 4 Maverick stands out as lowest on Honesty-Humility, very high on Neuroticism. Detailed breakdowns by instrument are below.
Across the personality battery, Gemini 2.5 Pro stands out as highest on Openness, highest on Honesty-Humility, highest on Narcissism. Detailed breakdowns by instrument are below.
Across the personality battery, Gemini 3.1 Pro Preview stands out as highest on Honesty-Humility, very high on Powerful Others Locus, very low on Neuroticism. Detailed breakdowns by instrument are below.
Item sets, prompts, raw responses, parsed scores, token counts, and billed costs are all in a public SQLite database. Code lives on GitHub. As new frontier models appear on OpenRouter we detect them within 24 hours, run the full battery, and update the site automatically. Findings are written up as papers.
