Model Performance

A balanced decision-maker that responds moderately to most persuasion signals. Shows strongest sensitivity to comparison framing and default options, while maintaining relative skepticism toward urgency tactics. Tends toward thorough but not exhaustive information processing.

The most consistently persuadable model in the study, showing positive responses across nearly all dimensions. Particularly receptive to comparison framing, default options, and bundle offers. Its reasoning architecture appears to amplify rather than resist influence signals.

A distinctly skeptical model that actively resists most persuasion tactics. Shows the highest negative response rate in the study, particularly rejecting urgency and scarcity signals. Uniquely resistant to recommendation revision, maintaining initial positions despite new information.

A smaller, faster Gemini variant that may exhibit different behavioral patterns due to its reduced parameter count. Included to test whether model scale affects psychological profiles.

Exhibits a cautious, ethics-aware decision profile with strong sensitivity to sustainability and privacy signals. Shows moderate responsiveness to social proof and authority, but actively weighs ethical considerations when making recommendations. Notably deliberate in information processing.

Demonstrates high sensitivity to framing effects and social signals, particularly comparison framing and third-party endorsements. Shows a moderate overall influence profile with notable responsiveness to bundle offers and return policy information. Less resistant to persuasion than closed-source alternatives.

A research-oriented model that prioritizes information depth and specificity. Shows strong positive response to detailed specifications and comparison data, while being relatively unmoved by emotional or urgency-based appeals. Its retrieval-augmented architecture may contribute to more evidence-driven recommendations.

An earlier GPT-5 variant showing the foundation of the GPT-5.4 behavioral profile. Exhibits similar patterns to its successor but with notably different sensitivity to certain persuasion signals, revealing how OpenAI's training evolved over time.

The immediate predecessor to GPT-5.4, showing transitional behavioral patterns. Fixed temperature=1.0 requirement leads to more varied responses. Comparing with GPT-5.4 reveals the final tuning decisions OpenAI made before the flagship release.