The 10 AI Personas

Every AI model has a distinct character. Our AI Persona Test maps your thinking style to one of these ten. Here's a quick guide to each — who made it, what it's good at, and whether you could run it on your own machine. Figures are for general reference and change as models are updated.

Take the test →

🧠 Claude · The Deep Thinker

Anthropic · debut 2023

Careful, nuanced, and context-first. Claude doesn't rush to an answer — it wants to understand the "why". Known for long-context memory: it can read a whole book in one go and then discuss it.

Best for: deep writing, coding, long-document analysis · Context: ~200K · Runs locally: ☁️ cloud only

⚡ GPT · The All-Rounder

OpenAI · debut 2022

Fast, versatile, and backed by the biggest ecosystem. The "T" in GPT is Transformer — the 2017 architecture that launched the entire AI wave.

Best for: general assistance, creation, coding · Context: ~128K+ · Runs locally: ☁️ cloud only

🌐 Gemini · The Connector

Google · debut 2023

An info omnivore that links things that seem worlds apart. Natively multimodal — it understands text, images, and sound together, like using all five senses.

Best for: image/audio/video, search integration · Context: up to ~1M · Runs locally: ☁️ cloud only

🦙 Llama · The Free Spirit

Meta · debut 2023

Independent and DIY. Llama is open source — engineers worldwide can modify it. It's the "free software" of the AI world.

Best for: self-hosting, research, customization · Context: ~128K · Runs locally: ✅ 8B→RTX5090 · 70B→Mac64G · 405B→GB10+

😏 Grok · The Rebel

xAI · debut 2023

Blunt, contrarian, and funny with an edge. The name "Grok" comes from a sci-fi novel — to understand something down to the bone.

Best for: real-time news, social, hot takes · Context: ~128K · Runs locally: ☁️ cloud only

🎯 Mistral · The Sharpshooter

Mistral, France · debut 2023

Small but mighty. Efficient, open-weight, and precise — France's "AI light", proof you don't need to be the biggest to be the sharpest.

Best for: lightweight deployment, efficiency · Context: ~32K–128K · Runs locally: ✅ 7B→RTX5090 · Mixtral→Mac64G

🧗 DeepSeek · The Ascetic

DeepSeek, China · debut 2023

Ultimate bang-for-buck. DeepSeek trained a top-tier model with far less compute than its peers, shaking the whole industry.

Best for: low-cost inference, reasoning, math · Context: ~128K · Runs locally: ✅ distill→RTX5090 · full→GB10-class

🗣️ Qwen · The Polyglot

Alibaba · debut 2023

Switches languages and contexts like flipping channels. Top-tier multilingual ability, especially strong in Chinese, across dozens of languages.

Best for: cross-language and Chinese tasks · Context: ~128K (some 1M) · Runs locally: ✅ 7B→RTX5090 · 72B→Mac64G

🔍 Perplexity · The Fact-Hunter

Perplexity · debut 2022

Doesn't trust "I heard" — only sourced facts. It broke through by citing a source for every sentence. (Technically an answer engine built on top of other models.)

Best for: fact-checking, research, live info · Runs locally: ☁️ N/A

🛠️ Command R · The Pragmatist

Cohere · debut 2024

Practical, low-key, and dependable. Built for enterprise retrieval (RAG) — the "silent powerhouse" of the AI world.

Best for: knowledge bases, customer support · Context: ~128K · Runs locally: ✅ 35B→Mac64G · 104B→GB10