Concepts
Models

AI Models

A model is the reasoning engine at the heart of every AI system. It's the part that reads your input and generates a response.

🧠
What is an AI Model?
A model is a mathematical system trained on vast amounts of text to predict what comes next — and in doing so, it learns to reason, write, and answer questions.
📖
Context window
How much text it can read in one go — up to 200,000 tokens (≈150,000 words)
Reasoning
Models can think step-by-step before answering — better accuracy on hard problems
💰
Speed vs cost
Smaller models are faster and cheaper; larger models are more capable
🎯
Instruction following
Modern models reliably do what you ask — the foundation of all agent behaviour
Agent Cookbook
0:00 / 0:07

What a model actually does

A language model is trained on enormous amounts of text — books, websites, code, conversations. Through this training it learns patterns, facts, reasoning strategies, and language itself.

When you send it a message, it doesn't "look up" an answer. It generates the most likely continuation of your prompt, word by word, based on everything it learned. The remarkable thing is that this simple process produces genuinely intelligent-seeming behavior at scale.

The key properties to understand

Context window

How much text the model can "see" at once — your prompt, its response, any documents you shared, and the conversation history. Measured in tokens (roughly 1 token ≈ ¾ of a word).

A small context window means the model forgets earlier parts of a long conversation. A large context window (100k+ tokens) means it can read an entire book or codebase in one shot.

Reasoning capability

Not all models are equally good at complex reasoning — breaking a problem into steps, catching logical errors, handling ambiguity. More capable models tend to be slower and more expensive.

Speed and cost

Smaller models are faster and cheaper. For high-volume, simple tasks (classifying emails, extracting data from forms), a smaller model often works fine and costs 10–100x less.

Instruction following

How reliably the model does what you ask. This varies significantly and is often more important than raw intelligence for production use cases.


The major models (2025–2026)

ModelMade byBest for
Claude 3.5 / 3.7AnthropicReasoning, writing, code, long documents, safety
GPT-4oOpenAIGeneral purpose, tool use, multimodal
Gemini 1.5 / 2.0GoogleVery long context, multimodal, Google ecosystem
Llama 3 / 4MetaOpen source, self-hosted, customizable
MistralMistral AIFast, efficient, European data compliance
DeepseekDeepseekStrong reasoning, lower cost

Beginner rule of thumb: Start with Claude for most tasks. It's excellent at following nuanced instructions, handles long documents well, and is designed with safety in mind. Switch to a different model only when you have a specific reason to.

Models vs. agents

A model alone is like a very smart person sitting in a room with no phone, no computer, no pen — just their mind. They can reason and advise brilliantly, but they can't do anything in the world.

An agent gives the model hands. It can browse the web, run code, send emails, update databases. That's the next page.

Next: What are agents? →