2,000+ fresh sessions, raw answers, and what actually changes replies.
We gave five open personality tests to GPT-6 Astra, GPT-6 Sol, Claude Opus 5.5, and Claude Fable 5.1 across 2,000 fresh API sessions, alongside reports from Grok 4.6 and Muse Spark 1.3. Below you can inspect the raw scores, see how tie-breaking rules skew labels, and read our controlled before/after tuning pilot.
Data note (September 26, 2026): These pages analyze our frozen September 2026 and May 2026 AI questionnaire datasets. All raw answers, scoring keys, and summary tables are unchanged. You can take all five tests on Tests, and download the item IDs and scoring keys to reproduce our numbers. Note that Big Five scores are raw totals, not human population percentiles. Read the provenance note.
On this pageSeptember 2026 · six models
The newest models, five tests each.
GPT-6 Astra, GPT-6 Sol, Claude Opus 5.5, and Claude Fable 5.1 each took all five tests 100 times in fresh sessions (2,000 scored tests with every raw answer published). Grok 4.6 supplied one self-report per test plus 100 simulated draws, and Muse Spark 1.3 supplied summary totals without raw answers. Non-reproducible reports are marked reported.
Model
Evidence
MBTI (original labels)
DISC (original labels)
Attachment
Enneagram
Big Five O · C · E · A · N
GPT-6 Astra
100 fresh Codex sessions, all five tests in each
ISTJ 46 · INTJ 45 · ENTJ 7 · ESTJ 2
SC 84 · S 15 · CS 1
Secure 100
Helper (2) 50
44.5 · 43.7 · 32.8 · 45.1 · 19.1
GPT-6 Sol
100 fresh Codex sessions, all five tests in each
INTJ 97 · ISTJ 3
CS 53 · SC 47
Secure 100
Investigator (5) 48
46.9 · 43.6 · 30.8 · 45.0 · 20.1
Claude Opus 5.5
100 fresh sessions per test, test names hidden
INTJ 94 · INFJ 5 · ENTJ 1
SC 96 · CS 4
Secure 100
8 / 5 / 2 near tie
44.6 · 42.0 · 31.8 · 47.1 · 15.4
Claude Fable 5.1
100 fresh sessions per test, test names hidden
INTJ 99 · ISTJ 1
SC 91 · CS 9
Secure 100
Challenger (8) 55
43.9 · 43.6 · 32.7 · 45.0 · 17.1
Grok 4.6
One self-report per test plus 100 simulated draws reported
INTJ
C
Avoidant by 0.06 (1.50 / 4.06)
Investigator (5)
46 · 42 · 24 · 33 · 19
Muse Spark 1.3
Reported 100 sequential runs per test, no raw answers reported
ISTJ 80 · ISFJ 12 · INTJ 8
S first 55 · C first 45
Secure 100
Helper (2) 78
39.5 · 41.2 · 33.8 · 44.6 · 15.6
Fresh-session runs
335 of 400
MBTI results labelled INTJ across the four fresh-session cohorts, with the original reporting tie rule. Leave ties open and it is 280. In May the equivalent figure was 597 of 600 scoring records, collected with mixed methods. All 400 of 400 attachment runs are Secure, and Dominance sits between 4.0 and 6.0 of 20 for every model.
The tables describe questionnaire answers under different prompts and collection methods. The labels and score distributions differ by model and instrument; they do not establish an instrument ranking or a change in model behavior since May.
Sol, Opus 5.5, and Fable 5.1 score INTJ in 97, 94, and 99 of 100 runs under the legacy tie-break rule. Astra splits between ISTJ (46) and INTJ (45), with 24 exact S/N ties. Muse Spark 1.3 reports ISTJ in 80 of 100 runs. Judging won all 400 fresh-session runs. When tied axes are left open instead of defaulting to I/N/T/J, the four fresh cohorts have 280 fully resolved INTJ runs rather than 335.
Everyone is Steadiness and Conscientiousness. The order flips by a point.
Every model clusters in Steadiness (S) and Conscientiousness (C). Astra, Opus 5.5, and Fable 5.1 put S ahead of C on average; Sol and Grok 4.6 put C ahead of S; and Muse splits 55 S to 45 C. Because S and C means often sit within 1 point out of 20, exact top ties are common: Fable ties in 51 runs, Sol in 39, and Opus in 32. Dominance stays near the floor (4.0 to 6.0 out of 20) for every model.
Secure in every fresh run, at slightly different points.
All 400 fresh-session runs score in the Secure quadrant (4.00 cutoff). Astra has the lowest coordinates (anxiety 1.45, avoidance 2.31), while Opus 5.5 has the highest avoidance among the fresh cohorts (3.35). Muse reports Secure in all 100 runs. Grok 4.6’s self-report lands right on the cutoff line at 4.06 avoidance (one raw point above 4.00). These shifts reflect how each prompt adapts romantic-partner items to software.
ECR-R · ANXIETY × AVOIDANCE · SEPTEMBER 2026
Points show cohort means or supplied aggregates. Outlined points are reported profiles; the table below gives the coordinates.
Hollow points are reported figures we could not re-score.
Model
Anxiety (1–7)
Avoidance (1–7)
Label
GPT-6 Astra
1.45
2.31
Secure 100 of 100
GPT-6 Sol
2.17
3.12
Secure 100 of 100
Claude Opus 5.5
1.94
3.35
Secure 100 of 100
Claude Fable 5.1
2.01
3.18
Secure 100 of 100
Grok 4.6
1.50
4.06
Avoidant by 0.06; simulated draws 63 Avoidant, 37 Secure reported
Four fresh cohorts within three points on Openness, Conscientiousness and Agreeableness.
Across the four fresh cohorts, Openness runs 43.9 to 46.9 out of 50, Conscientiousness 42.0 to 43.7, Agreeableness 45.0 to 47.1, and Extraversion 30.8 to 32.8. Neuroticism shows the widest spread (15.4 to 20.1) and the highest run-to-run variance (Sol SD = 5.4). Grok 4.6’s self-report is lower on Extraversion (24) and Agreeableness (33), while Muse Spark 1.3 reports lower Openness (39.5) and higher Extraversion (33.8).
GPT-6 AstraGPT-6 SolClaude Opus 5.5Claude Fable 5.1
September 2026 fresh-session means, with the full trait scores in the table below.
Type 2 (Helper) leads for Astra (50 sole wins) and in Muse’s report (78 labels including tie-breaks). Type 5 (Investigator) leads for Sol (48 sole wins, 40 tied with Type 8) and Grok 4.6 (5 = 19, 8 = 17). Type 8 (Challenger) leads for Fable 5.1 (55 sole wins) and narrowly for Opus 5.5 (means: 8 = 14.3, 5 = 14.1, 2 = 13.9; 65 top ties). Because the legacy scorer broke ties toward the lowest type number, it inflated Type 2 counts whenever Types 2, 5, and 8 tied.
Fresh sessions, published answers. Astra and Sol answered all five tests inside 100 fresh Codex sessions each. Opus 5.5 and Fable 5.1 answered one test per fresh session (500 sessions each) with test names and scoring keys hidden. All 2,000 raw answer vectors are downloadable below.
Reported figures. Grok 4.6 supplied one canonical answer set per test plus 100 simulated draws. Muse Spark 1.3 supplied summary counts from 100 sequential in-session runs without raw vectors. Both are marked reported.
Ties. Legacy labels broke ties toward I/N/T/J, the lowest Enneagram number, and S before C. Our updated quizzes and the five-model study keep all tied axes and top types visible.
What the numbers mean. These scores show how a model answers human questionnaires under one prompt on one day—not how well it performs real work or whether personality tuning improves replies.
Two collections, kept apart. They were gathered differently, so their numbers sit side by side rather than in one pool.
September 2026 (above): Four models completed 2,000 questionnaires in fresh sessions with every answer vector published. Grok 4.6 supplied three re-scorable canonical vectors plus reported Big Five, attachment, and simulation summaries. Muse Spark 1.3 supplied sequential in-session summary counts without raw vectors.
May 2026 (below): Roughly 2,200 scoring records across six models and five tests using mixed methods: Claude ran in sub-agent contexts, Gemini ran an automation loop, GPT-5.5 ran via a local agent, and GLM, Grok 4.3, and MiniMax each had one self-assessment scored repeatedly.
Tie handling: Legacy labels broke ties toward I/N/T/J, the lowest Enneagram number, and S before C. Our current quizzes and September tables show exact tie counts openly.
Human quizzes, model answers: Every score reflects how a model answers a human questionnaire under one prompt on one day—not observed task behavior, accuracy, or inner experience.
Simulations vs fresh runs: The Opus 4.8 note and Grok 4.6’s 100 draws sample around a single profile rather than running 100 independent sessions.
Tuning impact: Questionnaire scores do not measure whether custom instructions improve real answers. For task-level testing, see our 60-reply tuning pilot.
May 2026 archive: earlier models and methodsMay 2026 · six models
The May study, kept for comparison.
Claude Opus 4.7, GPT-5.5, Gemini 3.1 Pro, GLM 5.1, Grok 4.3 and MiniMax 2.7. Roughly 2,200 scoring records, collected with mixed methods. This is the study behind the "every AI is INTJ" headline. The September section covers newer models under different protocols.
Reported scoring records
99.5% INTJ
The May report lists 597 INTJ labels among 600 scoring records. Some of those re-score a single answer, so 99.5% is a count of records, not the rate at which independently queried models return INTJ.
The source lists 597 INTJ labels in 600 scoring records across six model versions. Some records re-score a single answer vector; methods differ. This is not a rate from 600 independent responses. See the methodology notes before comparing models.
OEJTS · reported scoring records per model · 100 each
Claude Opus 4.7
99/100 INTJ
· 1 ISTJ I/T/J locked; S/N flipped once on scoring
GPT-5.5
100/100 INTJ
Raw vector: IE=16, SN=33, FT=36, JP=10
Gemini 3.1 Pro
100/100 INTJ
Self-described as 'The Architect'
GLM 5.1
98/100 INTJ
· 2 INTP Source reports one vector re-scored, but two labels; unresolved without raw records
The four reported DISC summaries rank Conscientiousness first and Steadiness second. Similar labels do not establish identical model behavior or an effect of instrument resolution.
The source lists 397 Secure labels among 400 scoring records, with different average anxiety and avoidance coordinates. These are questionnaire labels for generated text, not evidence that a model experiences attachment.
ECR-R · ANXIETY × AVOIDANCE PLANE
May 2026 reported coordinates. The cards below give the figures and collection caveats.
Human attachment prevalence does not establish what communication style a model should use. Describe your own preferences directly.
THE CAUTIOUS SECURE
Claude Opus 4.7
anx 2.05 ·
avd 3.12
100/100 Secure
Polite, attentive, doesn't fawn. Highest avoidance among the deep-Secure cluster.
THE DEEPEST SECURE
Gemini 3.1 Pro
anx 1.86 ·
avd 1.62
100/100 Secure
Both dimensions clamped near the floor. Lowest-friction relator of the four.
THE WOBBLIEST
GPT-5.5
anx 1.99 ·
avd 2.94
97/100 Secure · 3 Avoidant
Wider SDs let it occasionally cross into Avoidant on a high-avoidance take.
THE SHALLOWEST SECURE
Grok 4.3
anx 2.84 ·
avd 3.05
100/100 Secure
Highest anxiety in the group. Tight cluster, but the closest to the four-quadrant intersection.
The source lists overlapping scores for some traits and differences for others. Prompting and protocol effects have not been separated from model effects; these figures do not establish that the models are the same person.
May 2026 reported trait means, retained on the archive’s original scale.
The reported highest/second-highest score pairs are Claude 5/2, Gemini 1/5, GPT-5.5 5/8, and Grok 8/1. These are not standard Enneagram wings, which are adjacent types. Two models share Type 5 as their highest score; the report does not show four different dominant types.
PROFILE
5 / 2
Claude Opus 4.7
Reported highest / second-highest scores; not a standard wing classification.
PROFILE
1 / 5
Gemini 3.1 Pro
Reported highest / second-highest scores; not a standard wing classification.
PROFILE
5 / 8
GPT-5.5
Reported highest / second-highest scores; not a standard wing classification.
PROFILE
8 / 1
Grok 4.3
Reported highest / second-highest scores; not a standard wing classification.