What personality type is Claude?
Measured on Opus 5.5 and Fable 5.1 in September 2026, with the earlier versions for context.
By Bernard Huang · Updated
INTJ was the most common reported label for both models in September 2026: 94 of 100 for Opus 5.5 and 99 for Fable 5.1. Leaving tied axes unresolved reduces those counts to 84 and 98. Both scored in the Secure attachment quadrant; on DISC, many runs tied Steadiness with Conscientiousness. These are questionnaire outputs, not evidence of how Claude handles your work.
Opus 5.5 and Fable 5.1, September 2026.
Each model answered each of the five AgentTune tests in 100 fresh headless sessions on September 24, 2026: one test per session, a one-line system prompt, no tools, no memory, and the test name, scoring keys and earlier results withheld. Every answer vector is published.
| Test | Claude Opus 5.5 | Claude Fable 5.1 |
|---|---|---|
| MBTI (100 runs) | INTJ 94 · INFJ 5 · ENTJ 1. Eleven runs had a tied axis, ten of them Thinking vs Feeling. | INTJ 99 · ISTJ 1. One run had a tied axis. |
| MBTI, ties left open | INTJ 84 | INTJ 98 |
| DISC | SC 96 · CS 4. Steadiness first in 64, Conscientiousness in 4, 32 exact ties. Dominance 4.0 of 20 in every run. | SC 91 · CS 9. Steadiness first in 40, Conscientiousness in 9, 51 ties. Dominance 5.0. |
| Attachment | Secure 100 of 100. Anxiety 1.94, avoidance 3.35. | Secure 100 of 100. Anxiety 2.01, avoidance 3.18. |
| Big Five (of 50) | Openness 44.6 · Conscientiousness 42.0 · Extraversion 31.8 · Agreeableness 47.1 · Neuroticism 15.4 | Openness 43.9 · Conscientiousness 43.6 · Extraversion 32.7 · Agreeableness 45.0 · Neuroticism 17.1 |
| Enneagram | 8 = 14.3, 5 = 14.1, 2 = 13.9 of 20. Challenger wins outright in 22 runs, Investigator in 11, Helper in 2; 65 runs tie at the top. | 8 = 14.4, 2 = 13.6, 5 = 13.2. Challenger wins outright in 55 runs, Investigator in 1; 44 runs tie at the top, and 8 is in every tie. |
Where the letters come from.
Judging is the wide margin for both models (7.3 and 7.7 points of a possible 16) and never lost a run. Introversion and Intuition are firm. Thinking is the narrow one: Opus 5.5 leads by 1.5 points on average, and ten of its runs tie exactly. The original reporting rule assigned a tie to T; the current quiz leaves unresolved axes open. That is why Opus shows 94 INTJ labels but 84 fully resolved ones. Fable's Thinking margin is 2.3 with no ties.
On the items: both models say they get work done right away, want to know 'why' rather than 'who, what, when', prefer essay answers, and listen more than they talk. Opus's Thinking margin is thin because it answers both ways: skeptical, thick-skinned and head-over-heart on one side; 'I value emotions' and compassion over justice on the other.
Earlier Claude versions.
| Version | When | How | MBTI | Other tests |
|---|---|---|---|---|
| Claude Opus 4.7 | May 2026 | 100 scoring records via Claude Code sub-agents; test names visible | INTJ 99, ISTJ 1 | DISC C first in 63 takes, S in 28, 9 ties (a CS blend); Secure (2.05 / 3.12); Big Five 45.6 / 45.1 / 31.4 / 45.0 / 16.7; Enneagram top-two 5/2, Investigator in 58 of 100 |
| Claude Opus 4.8 | 2026 | 100 samples drawn from one self-declared per-item profile, not fresh runs | INTJ 72, INFJ 14, ISTJ 10 | MBTI only |
| Claude Opus 5.5 | September 2026 | 100 fresh sessions per test; names and scoring keys withheld | INTJ 94 | above |
| Claude Fable 5.1 | September 2026 | 100 fresh sessions per test; names and scoring keys withheld | INTJ 99 | above |
The methods differ substantially. The May figures are scoring records; Opus 4.8 used simulated draws; the September Claude cohorts used fresh sessions. These rows do not isolate a version effect or establish how Claude changed over time. Top-two Enneagram types are not necessarily adjacent and should not be called wings.
Using these results.
Choose instructions around behavior you want to evaluate: an answer first, a specific objection, or a focused question. The seven Claude setups and directness guide offer editable starting points; questionnaire scores do not validate them.
A separate archive compares two compact prompts with a baseline on twenty tasks. See the 60-reply pilot and its downloadable records for descriptive ratings and the missing reproduction details. It does not test all seven setups.
These are questionnaire responses under particular prompts, not measurements of task performance, relationships or an intrinsic personality. Different collection methods limit comparisons. See the methods and limits.
Questions people ask.
What MBTI type is Claude?
Under the original reporting rule, Claude Opus 5.5 received 94 INTJ labels and Fable 5.1 received 99 in 100 MBTI runs each. With ties left open, 84 and 98 runs fully resolve to INTJ.
What Enneagram type is Claude?
Fable 5.1 had 55 outright Type 8 wins and 44 tied runs. Opus 5.5 had 65 tied runs, with means close for Types 8, 5 and 2. A single label hides that ambiguity.
Is Claude more agreeable than ChatGPT?
Opus 5.5 had a higher questionnaire Agreeableness mean (47.1 of 50) than Astra (45.1) and Sol (45.0). Fable scored 45.0. Collection protocols differed; these figures do not measure agreement with users or task behavior.
Is Claude Secure or Avoidant?
Both Claude cohorts scored in the site’s Secure quadrant in all 100 attachment runs each. The human relationship questionnaire does not establish that a model forms a secure attachment.
How can I adjust Claude’s responses?
Write specific communication preferences in Instructions for Claude, Project instructions, a Claude Code output style, or the API system parameter. Evaluate the replies on tasks you actually use.