• •Models: Claude Opus, Sonnet, Haiku
• •Best for: Long-context tasks, coding, analysis, safety-conscious applications
• •Context window: Up to 1 million tokens
• •Pricing: Pay per token (input and output priced separately)

• •Models: GPT-4o, GPT-4, GPT-3.5 Turbo
• •Best for: General-purpose tasks, wide ecosystem, plugins
• •Context window: Up to 128K tokens
• •Pricing: Pay per token

• •Models: Gemini Ultra, Pro, Flash
• •Best for: Multimodal tasks (text + images), Google ecosystem integration
• •Context window: Up to 1 million tokens
• •Pricing: Free tier available, then pay per token

Sign up at the Anthropic Console. Create a new API key. Keep it secret -- never put API keys in your code directly.

> pip install anthropic

Here is the simplest possible example:

> import anthropic

>

> client = anthropic.Anthropic(api_key="your-api-key-here")

>

> message = client.messages.create(

> model="claude-sonnet-4-6-20250514",

> max_tokens=1024,

> messages=[

> {"role": "user", "content": "Explain quantum computing in 3 sentences."}

> ]

> )

>

> print(message.content[0].text)

That is it. Run this and you will get a response from Claude explaining quantum computing.

AI models process text in chunks called tokens. A token is roughly 4 characters or 3/4 of a word.

• •"Hello, how are you?" = about 6 tokens
• •A typical email = 200-500 tokens
• •A full blog post = 1,000-3,000 tokens

You pay for both input tokens (what you send) and output tokens (what you get back). Output tokens are usually more expensive.

The context window is the maximum amount of text the model can consider at once. Think of it as the model's working memory.

| Model | Context Window | Roughly Equivalent To |

| --- | --- | --- |

| GPT-3.5 Turbo | 16K tokens | ~25 pages |

| GPT-4o | 128K tokens | ~200 pages |

| Claude Sonnet | 200K tokens | ~300 pages |

| Claude Opus | 1M tokens | ~1,500 pages |

| Gemini Pro | 1M tokens | ~1,500 pages |

Temperature controls how creative or predictable the model's responses are:

• •Temperature 0: Very predictable, always picks the most likely response. Good for factual tasks.
• •Temperature 0.5: Balanced. Good for most applications.
• •Temperature 1.0: More creative and varied. Good for brainstorming and creative writing.

A system prompt tells the model how to behave. It is like giving instructions to a new employee:

> messages = [

> {"role": "user", "content": "What should I have for dinner?"}

> ]

Without a system prompt, you get a generic response. With one:

> system = "You are a nutritionist who specializes in quick, healthy meals for busy professionals. Keep responses under 100 words."

Now the model responds as a nutritionist, not a generic assistant.

To have a multi-turn conversation, include the full message history:

> messages = [

> {"role": "user", "content": "I am building a todo app in React."},

> {"role": "assistant", "content": "Great choice! React is perfect for a todo app..."},

> {"role": "user", "content": "How should I structure the components?"}

> ]

The model uses all previous messages as context for its response.

For a better user experience, you can stream responses token by token instead of waiting for the full response:

> with client.messages.stream(

> model="claude-sonnet-4-6-20250514",

> max_tokens=1024,

> messages=[{"role": "user", "content": "Write a short story about a robot."}]

> ) as stream:

> for text in stream.text_stream:

> print(text, end="", flush=True)

This gives you the "typing" effect you see in ChatGPT and Claude.

1. 1AI-powered FAQ bot -- Feed it your documentation, let it answer questions
2. 2Content summarizer -- Paste in long articles, get concise summaries
3. 3Code explainer -- Paste code, get plain-English explanations

1. 1Email draft assistant -- Generate professional email responses from bullet points
2. 2Study flashcard generator -- Feed it textbook content, get flashcards
3. 3Recipe generator -- Input ingredients you have, get recipe suggestions

1. 1AI customer support -- Integrate with your product's knowledge base
2. 2Document analyzer -- Upload PDFs and ask questions about them
3. 3Code review bot -- Automatically review pull requests and suggest improvements