Beginner-Friendly Topic
Take your time - it's perfectly normal to re-read this topic 2-3 times. Try the interactive code editor below to run code yourself. Use the Q&A section to check your understanding before moving on. You've got this! 🚀
AI Models
An AI model is the result of the training process. It is a mathematical structure that stores everything the AI learned from the training data. When you use ChatGPT or Gemini, you are interacting with a model.
What is an AI Model?
An AI model is a large mathematical function. It takes an input (your prompt) and produces an output (the AI's response).
Internally, a model consists of:
- Layers: organized sections of computation
- Parameters (weights): numbers that store the learned patterns
- A process for converting input to output
A small AI model might have a few hundred million parameters. A large one can have hundreds of billions. Each parameter is just a number, but together they encode everything the model learned.
You do not need to understand the mathematics. Just know that 'the model' is the trained AI system that responds to your prompts.
Deep Learning ⊂ Machine Learning ⊂ Artificial Intelligence
Common AI Models You May Have Heard Of
- GPT-4 and GPT-4o: OpenAI's models that power ChatGPT and the OpenAI API
- Gemini 1.5 Pro and Gemini 2.0 Flash: Google's models that power Google Gemini and Vertex AI
- Claude 3.5 and Claude 4: Anthropic's models known for being careful and safety-focused
- Llama 3: Meta's open-source model that anyone can download and run locally
- Mistral: European open-source models known for efficiency
- Stable Diffusion: open-source image generation model
- DALL-E 3: OpenAI's image generation model used in ChatGPT
Common Mistake
Warning
People often confuse the model with the application. ChatGPT is an application built by OpenAI. It uses the GPT model underneath. The model is the AI brain. The application is the interface you use to talk to it. You can access the same GPT model through the OpenAI API and build your own applications with it.
Key Takeaways
- An AI model is the result of the training process.
- GPT-4 and GPT-4o: OpenAI's models that power ChatGPT and the OpenAI API
- Gemini 1.5 Pro and Gemini 2.0 Flash: Google's models that power Google Gemini and Vertex AI
- Claude 3.5 and Claude 4: Anthropic's models known for being careful and safety-focused