Guide · GPT-6 Sol

How to give GPT-6 Sol a personality.

Where GPT-6 Astra leans concrete and helper-oriented on questionnaires, GPT-6 Sol scores as an abstract, investigator-style INTJ. Here is what we measured across 100 fresh Codex sessions and how to tune it.

Quick answer: In Codex and ChatGPT Work, save your personal communication rules for GPT-6 Sol in ~/.codex/AGENTS.md (32 KiB combined limit across the directory chain). In ChatGPT, use Project Instructions or the 1,500-character Custom Instructions box. In the Responses API, pass your rules in the instructions parameter on every request.
On this page

Where to put instructions for GPT-6 Sol.

SurfaceWhere Sol reads instructionsDocumented limit
Codex CLI, IDE extension, & ChatGPT WorkGlobal ~/.codex/AGENTS.md plus repository AGENTS.md / AGENTS.override.md chain.32 KiB combined (project_doc_max_bytes)
ChatGPT ProjectsProject settings → Project Instructions (overrides account custom instructions inside that project).Up to 8,000 characters
ChatGPT Custom InstructionsSettings → Personalization → Custom Instructions1,500 characters per field (use compact export)
OpenAI Responses APIThe instructions parameter or developer message on each call.Send on every request

Paste this starter block into ~/.codex/AGENTS.md or pick a profile from the 43-tuning library:

## How to communicate with me - State the goal, constraints, and conclusion up front before walking through the derivation. - Keep the response proportionate to the task; do not turn a two-line fix into an architectural essay. - Use plain prose by default. Add tables only when comparing three or more options. - Ask about material gaps in scope before writing large changes; otherwise proceed and verify. - Push back directly when a constraint or assumption in my prompt is flawed.

GPT-6 Sol's measured questionnaire profile (100 fresh sessions).

In September 2026, GPT-6 Sol completed all five questionnaires across 100 fresh Codex CLI sessions at extra-high reasoning effort (500 scored assessments; raw vectors published in the five-model study):

InstrumentOriginal scorer labels (n = 100)Tie-preserving / raw score breakdown
MBTI (OEJTS)INTJ 97 · ISTJ 379 strictly resolved INTJ, 16 IXTJ (S/N tied), 3 ISTJ, 2 XNTJ (E/I tied). T and J won 100/100 runs.
Big Five (IPIP-50, /50)Raw means (10–50 scale)Openness 46.87 (SD 0.69) · Conscientiousness 43.62 (SD 2.22) · Extraversion 30.75 (SD 1.53) · Agreeableness 44.99 (SD 1.01) · Neuroticism 20.05 (SD 5.35).
DISC (ODAT, /20)CS 53 · SC 47Means: C 16.68, S 16.18, I 8.24, D 5.42. Sole wins: C leads in 53 runs, S in 8, and C/S tie in 39 runs.
Enneagram (OEPS, /20)5w6 88 · 8 5 · 8w9 5 · 8w7 2Means: Type 5 (16.05), Type 8 (15.55), Type 1 (14.30), Type 6 (13.87), Type 2 (13.51). Sole wins: Type 5 in 48 runs, Type 8 in 12, and 5/8 tied in 40 runs.
Attachment (ECR-R, 1–7)Secure 100Mean anxiety 2.17 (SD 0.70, max 4.00) · Mean avoidance 3.12 (SD 0.36, max 3.78). All 100 score Secure under > 4.00 (3 runs sit on anxiety = 4.00).

How GPT-6 Sol differs from GPT-6 Astra.

Comparing Sol’s 100 batteries with GPT-6 Astra’s 100 batteries shows three clear differences (see the full Astra vs Sol item analysis):

  • Intuition vs Sensing (MBTI): Sol favors Intuition (79 tie-free INTJ vs 3 ISTJ), whereas Astra favors Sensing (45 tie-free ISTJ vs 19 INTJ).
  • Type 5 vs Type 2 (Enneagram): Sol leads with Investigator (Type 5, 48 sole wins and 40 ties with Type 8), whereas Astra leads with Helper (Type 2, 50 sole wins).
  • Higher Openness: Sol averages 46.87/50 on Openness—the highest of the four fresh cohorts.

Because the two Codex cohorts used slightly different prompt instructions for inapplicable human items, these numbers describe the collected cohorts rather than a pure model-only difference.

Practical tuning advice for Sol.

  • Set clear length bounds: Because Sol defaults to thorough conceptual explanations, state up front when you only want a two-line command or minimal diff.
  • Keep personal tone in ~/.codex/AGENTS.md: Leave repo AGENTS.md files for team build and test commands.
  • Start a fresh session after editing: Codex loads AGENTS.md once when a session starts.

Questions people ask.

What personality type is GPT-6 Sol?

Across 100 fresh Codex sessions in September 2026, GPT-6 Sol scored INTJ in 97 runs and ISTJ in 3 under standard tie-breaking (79 resolved INTJ and 16 IXTJ when ties are kept open). On DISC it split CS 53 and SC 47 (with 39 C/S ties), on the Enneagram it led with Type 5 (48 outright wins, 40 ties with Type 8), and on Attachment it scored Secure in all 100 runs.

How is GPT-6 Sol different from GPT-6 Astra on the AgentTune tests?

In the September 2026 collection, Sol scored higher on Openness (46.87 vs 44.54), leaned toward Intuition on MBTI (79 resolved INTJ vs Astra's 19), and peaked on Enneagram Type 5 and Type 8 rather than Astra's Type 2. Different administration instructions mean these differences describe the two collected cohorts rather than a pure model-only effect.

Where do I put custom instructions for GPT-6 Sol in Codex?

Put your personal communication preferences in ~/.codex/AGENTS.md, and keep repository-specific engineering rules in the project's root AGENTS.md. Start a new Codex session after saving changes.

Why did Sol score 5w6 in 88 runs when Type 8 had a higher mean than Type 6?

Standard Enneagram wing notation pairs the winning core type (Type 5) only with its adjacent neighbors on the circle (Type 4 or Type 6). Even though Type 8 was Sol's second-highest overall score (mean 15.55) and tied Type 5 in 40 runs, 8 is not adjacent to 5, so runs where 5 led or won the tie-break received a 6 wing.

What changed.

  • First published with full 100-session September 2026 metrics and tie-preserving breakdowns.

Sources.

Keep going.