GenAI Cheatsheet

🧾 GenAI Cheatsheet

🔠 LLM (Large Language Model)

• Focus: Text (answering, writing, summarizing, translating).
• Examples: GPT, Llama, Claude.

🎨 Generative Models

• Diffusion models → images & video (Stable Diffusion, DALL·E, Sora).
• Audio models → voice & music (Suno, MusicLM).

🌐 GenAI (Generative AI)

• Umbrella term → all models that generate content.
• Subgroups:
  • Text → LLMs
  • Images → diffusion
  • Video → diffusion / 3D transformers
  • Audio → voice + music
  • Multimodal → mix of modalities

🔀 Multimodal

• Definition: Models that understand + generate across text, images, audio, video.
• Examples: GPT-4o, Gemini, Claude 3.5, Kosmos.

🔌 MCP (Model Context Protocol)

• What it is: A protocol for LLMs to connect with external apps (APIs, databases, Gmail, Notion…).
• Benefits:
  • Dynamic discovery of capabilities
  • Simple, secure integration
  • LLM doesn’t need to “know Gmail” → it just knows MCP