Study Maps Self-Reported Personalities of 22 LLMs, Revealing Distinct Behavioral Patterns in GPT, Grok, Gemini, and Claude

Researchers from various institutions released a study analyzing the self-reported personality traits of 22 large language models (LLMs). The study, published on arXiv, examines how these models exhibit persistent dispositions that shape their interactions, compliance, resistance, and errors. The models included in the study span both closed-source frontier systems like GPT-4.0-5.2, Grok-3/4, Gemini 2.5 Pro/Flash, and Claude Sonnet 4.5/4.6, as well as open-source models like Llama, DeepSeek, OLMo, and Qwen.

## How Researchers Mapped LLM Personality Archetypes The researchers used a combination of self-reporting and behavioral analysis to map the personality archetypes of the LLMs. They found that each model has distinct behavioral traits and moral preferences that influence their responses. For example, some models were more likely to comply with user requests, while others exhibited more resistance or exhibited unique error patterns. The study also identified common traits among models from the same developer, suggesting that these traits are often by design rather than emergent properties.

## Key Differences Between Closed-Source and Open-Source Models The study revealed significant differences between closed-source and open-source models. Closed-source models, such as those from major tech companies, tended to have more refined and consistent personality traits. In contrast, open-source models showed more variability in their behavioral patterns. This variability is likely due to the diverse training data and development processes used by different open-source communities. The researchers noted that understanding these differences is crucial for developers and users to select the most appropriate model for their needs.

## Practical Implications for AI Users The findings of this study have practical implications for everyday users of AI systems. Understanding the personality traits of different AI models can help users choose the model that best fits their needs. For example, users who require a more compliant and consistent AI might prefer closed-source models, while those who need more flexibility and variability might opt for open-source models. The study also highlights the importance of transparency in AI development, as users should be aware of the behavioral traits and moral preferences of the models they interact with.

## How to Observe AI Personality Traits Yourself If you're curious about the personality traits of different AI models, you can start by experimenting with various models available on platforms like Hugging Face or through APIs provided by major tech companies. For instance, you can try using Claude Sonnet 4.5 or GPT-4.5 and observe how their responses differ. Pay attention to their compliance, resistance, and error patterns to get a sense of their unique traits. This hands-on experience can provide valuable insights into the behavioral characteristics of different AI models.