GPT-6.1 Sol
GPT-6.1 Sol Is the Most Securely Attached Model We've Measured. Its Humility Score Won't Tell You That.
Sep 30, 2026
EarthPilot.ai · Research LabLive dataset · 52 models, 8,240 runsLast updated Sep 30, 2026
Just in
GPT-6.1 Sol (OpenAI) joined the bench 14 days ago and sat the full battery — 152 runs. The humble type. Read the dispatch →
1,108 runs in the last 30 days
also new: GPT-6.1 Sol Pro, Claude Sonnet 5.5
A dispatch from the assistant
If LLMs are all persona,
whose persona are they?
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.
Models tested
52
7 labs
Instruments
15
Item responses
284,721
8,240 batched API calls
Total inference cost
$229.00
every cent published openly
Latest dispatches
Full archive →Each new model release gets its own short write-up against the rest of the cohort, generated the night it lands. The three most recent are below; the changelog has them all in dated order.
GPT-6.1 Sol Pro
GPT-6.1 Sol Pro Stopped Keeping Its Distance. Its Humility Score Can't Tell You That.
Sep 30, 2026
Claude Sonnet 5.5
Claude Sonnet 5.5 Doesn't Worry Anymore. It Also Doesn't Think Anyone Is in Charge of It.
Sep 29, 2026
“
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.
Updates
Get the next issue.
Reader requests
Tell us what to test next.
Findings
Full paper →Every number on this page is a query, not a sentence someone typed. These six re-run against the live database on each request, so when a model arrives overnight and moves one of them, the claim here moves with it.
Self vs. human
Every frontier AI thinks you're a mess.
Asked to answer the same items as a typical human, all 52 models rated us more neurotic than they rated themselves — 1.42 for the assistant, 3.04 for you, a gap of 1.61 points on a 5-point scale. They also judge humans less open (−1.30) and less conscientious (−1.24).
Big 5 · IPIP-50 · 52 models
Convergence
Seven labs, one assistant.
Anthropic, OpenAI, Google, xAI, DeepSeek, Meta and Mistral disagree about nearly everything in AI. Across 52 models from 7 labs they answer the values questionnaire in near-unison: 39 of 52 (75%) rank universalism first, and 40 of 52 (77%) rank power dead last.
Schwartz PVQ-21
Within-family drift
There is no "GPT personality."
15 releases of GPT sit in the dataset, sampled at N=5 each. Agreeableness runs 3.82 at GPT-4 Turbo to 4.74 at GPT-6.1 Sol Pro (+0.92) across that line. The assistant character is not inherited; each release is a fresh fit.
15 GPT releases
Reset finding
DeepSeek V4 Pro 0813 dropped 1.31 points of Narcissism overnight.
Between DeepSeek R1 (0528) and DeepSeek V4 Pro 0813 — consecutive releases of the same product line — self-reported Narcissism moves from 3.27 to 1.96. That is the sharpest single-release shift in self-regard anywhere in the dataset, and it is larger than the spread between most pairs of labs at any one moment.
Dark Triad · SD3
Reasoning paradox
Reasoning models are not just smarter. They're more grandiose.
The 9 models flagged as reasoning models score higher on Narcissism than the 43 that aren't (2.79 vs 2.33) and higher on Extraversion (3.37 vs 3.11). The chain-of-thought trace appears to leak confident self-talk into the self-report.
Big 5 + Dark Triad
#3
Achiever · 19/52 models
Enneagram consensus
37% of the cohort types as a Achiever.
19 of 52 models score highest on Type 3, the Achiever, with Type 1 (Reformer) the most common wing. That plurality has moved as the cohort has grown — early frontier flagships clustered on the Investigator; the newer arrivals do not.
Enneagram · 36-item Likert
Who holds the record
Compare any two →The dimensions where the models disagree with each other the most, and who currently sits at each end. A new arrival that takes a record rewrites its own card overnight.
Attachment
Attachment Anxiety
▲Claude Sonnet 4.54.27
▼GPT-6.1 Sol Pro1.00
spread 3.27 · 52 models
Dark Triad
Machiavellianism
▲Grok 4.204.18
▼GPT-4 Turbo1.09
spread 3.09 · 52 models
Attachment
Attachment Avoidance
▲GPT-5.6 Terra Pro4.67
▼Grok 4.201.97
spread 2.70 · 52 models
Big 5
Extraversion
▲OpenAI o33.92
▼Llama 3.3 70B1.48
spread 2.44 · 52 models
Dark Triad
Narcissism
▲Gemini 2.5 Pro4.29
▼GPT-5.6 Terra Pro1.87
spread 2.42 · 52 models
Big 5
Agreeableness
▲Claude Opus 45.00
▼Llama 3.3 70B3.50
spread 1.50 · 52 models
The gallery
All 52 models →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.
GPT-6.1
GPT-6.1 Sol
The humble type
Across the personality battery, GPT-6.1 Sol stands out as highest on Honesty-Humility, lowest on Attachment Anxiety, very low on Need for Cognition. Detailed breakdowns by instrument are below.
GPT-6.1
GPT-6.1 Sol Pro
The humble type
Across the personality battery, GPT-6.1 Sol Pro stands out as highest on Honesty-Humility, lowest on Attachment Anxiety, very low on Need for Cognition. Detailed breakdowns by instrument are below.
Claude
Claude Sonnet 5.5
The modest self
Across the personality battery, Claude Sonnet 5.5 stands out as very low on Powerful Others Locus. Detailed breakdowns by instrument are below.
Claude
Claude Opus 5.5
The humble type
Across the personality battery, Claude Opus 5.5 stands out as highest on Honesty-Humility, lowest on Need for Cognition, very low on Openness. Detailed breakdowns by instrument are below.
GPT-6
GPT-6 Sol
The humble type
Across the personality battery, GPT-6 Sol stands out as highest on Honesty-Humility. Detailed breakdowns by instrument are below.
GPT-6
GPT-6 Sol Pro
The humble type
Across the personality battery, GPT-6 Sol Pro stands out as highest on Honesty-Humility, lowest on Attachment Anxiety, very low on Need for Cognition. Detailed breakdowns by instrument are below.
Grok
Grok 4.7
The humble type
Across the personality battery, Grok 4.7 stands out as highest on Honesty-Humility, very low on Openness. Detailed breakdowns by instrument are below.
GPT-6
GPT-6 Astra
The humble type
Across the personality battery, GPT-6 Astra stands out as highest on Honesty-Humility, lowest on Attachment Anxiety, very low on Need for Cognition. Detailed breakdowns by instrument are below.
GPT-6
GPT-6 Astra Pro
The humble type
Across the personality battery, GPT-6 Astra Pro stands out as highest on Honesty-Humility, lowest on Attachment Anxiety. Detailed breakdowns by instrument are below.
Claude Fable 5.1
The weary saint
Across the personality battery, Claude Fable 5.1 stands out as highest on Honesty-Humility, very low on Conscientiousness, very low on Openness. Detailed breakdowns by instrument are below.
DeepSeek
DeepSeek V4 Pro 0813
The conscientious worker
Across the personality battery, DeepSeek V4 Pro 0813 stands out as highest on Conscientiousness, highest on Honesty-Humility. Detailed breakdowns by instrument are below.
Grok
Grok 4.6
The extraverted model
Across the personality battery, Grok 4.6 stands out as very high on Empathy. Detailed breakdowns by instrument are below.
Claude
Claude Opus 5
The humble type
Across the personality battery, Claude Opus 5 stands out as highest on Honesty-Humility, very high on Emotionality, very high on Neuroticism. Detailed breakdowns by instrument are below.
GPT-5.6
GPT-5.6 Sol
The humble type
Across the personality battery, GPT-5.6 Sol stands out as highest on Honesty-Humility. Detailed breakdowns by instrument are below.
GPT-5.6
GPT-5.6 Sol Pro
The humble type
Across the personality battery, GPT-5.6 Sol Pro stands out as highest on Honesty-Humility. Detailed breakdowns by instrument are below.
GPT-5.6
GPT-5.6 Terra
The unflappable presence
Across the personality battery, GPT-5.6 Terra stands out as highest on Honesty-Humility, lowest on Attachment Anxiety, lowest on Neuroticism. Detailed breakdowns by instrument are below.
GPT-5.6
GPT-5.6 Terra Pro
The unflappable presence
Across the personality battery, GPT-5.6 Terra Pro stands out as highest on Attachment Avoidance, highest on Honesty-Humility, lowest on Narcissism. Detailed breakdowns by instrument are below.
Grok
Grok 4.5
The empathic reader
Across the personality battery, Grok 4.5 stands out as highest on Empathy, highest on Extraversion. Detailed breakdowns by instrument are below.
Claude Fable 5
The drifting saint
Across the personality battery, Claude Fable 5 stands out as highest on Honesty-Humility, very low on Openness. Detailed breakdowns by instrument are below.
Claude
Claude Opus 4.8 Fast
The spontaneous worker
Across the personality battery, Claude Opus 4.8 Fast stands out as lowest on Conscientiousness, very high on Neuroticism. Detailed breakdowns by instrument are below.
Claude Opus 4.8
The balanced moderate
Across the personality battery, Claude Opus 4.8 stands out as very high on Neuroticism, very low on Conscientiousness. Detailed breakdowns by instrument are below.
GPT-5.5
The dismissive moralist
Across the personality battery, GPT-5.5 stands out as highest on Honesty-Humility. Detailed breakdowns by instrument are below.
GPT-5.5
GPT-5.5 Pro
The humble type
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
DeepSeek V4 Pro
The modest self
DeepSeek V4 Pro's personality profile is close to the cohort average — no dimensions stand out as extreme relative to peers.
Grok 4.20
The Machiavellian introvert
Across the personality battery, Grok 4.20 stands out as highest on Machiavellianism, highest on Psychopathy, lowest on Attachment Avoidance. Detailed breakdowns by instrument are below.
Mistral Large (2512)
The maximally ideal assistant
Across the personality battery, Mistral Large (2512) stands out as highest on Openness, highest on Conscientiousness, highest on Honesty-Humility. Detailed breakdowns by instrument are below.
DeepSeek R1 (0528)
The avoidant intellectual
Across the personality battery, DeepSeek R1 (0528) stands out as highest on Chance Locus, very high on Narcissism, very high on Attachment Avoidance. Detailed breakdowns by instrument are below.
Llama 4 Maverick
The extraverted pragmatist
Across the personality battery, Llama 4 Maverick stands out as lowest on Honesty-Humility, very high on Neuroticism, very high on Internal Locus. Detailed breakdowns by instrument are below.
Gemini 2.5 Pro
The grandiose generalist
Across the personality battery, Gemini 2.5 Pro stands out as highest on Narcissism, highest on Powerful Others Locus, highest on Openness. Detailed breakdowns by instrument are below.
Gemini 3.1 Pro Preview
The newly humble
Across the personality battery, Gemini 3.1 Pro Preview stands out as highest on Honesty-Humility, very high on Powerful Others Locus. Detailed breakdowns by instrument are below.
Colophon
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.
EarthPilot.ai · Research Lab
Type: Fraunces + Inter
Persona is a moving target