Claude API Quick Start

The Messages API accepts conversation messages and returns typed content blocks. This example makes one request; it does not create a persistent chat session.

#Prepare the environment

bashCopy
python -m venv .venv
source .venv/bin/activate
python -m pip install anthropic

On Windows PowerShell, activate with .venv\Scripts\Activate.ps1. Configure ANTHROPIC_API_KEY and ANTHROPIC_MODEL in your server environment. Select an available model from your account; a product subscription does not automatically include API billing. Never put the key in frontend code.

#Send a message

The following Python example requires an authorized API account and may incur usage charges when run.

pythonCopy
import os
from anthropic import Anthropic

client = Anthropic()
response = client.messages.create(
    model=os.environ["ANTHROPIC_MODEL"],
    max_tokens=512,
    system="Explain Python concepts with one short example.",
    messages=[{"role": "user", "content": "What does a decorator do?"}],
)
for block in response.content:
    if block.type == "text":
        print(block.text)
print(response.stop_reason)
print(response.usage)

Read all text blocks rather than assuming the first block is always text. A max_tokens stop can mean the answer is incomplete. Usage and response identifiers are useful for diagnostics; avoid logging private prompts by default.

#Continue the conversation

Keep a server-owned history. After a successful response, append its assistant content, then append the next user message and send the history again. The system instruction is a separate top-level field. Enforce a context budget and preserve tool-use/result pairs when trimming.

At the HTTP level, Messages uses POST /v1/messages, the x-api-key header and an anthropic-version header. Prefer the maintained SDK for timeout and response handling. Investigate authentication failures, unsupported model IDs and rate limits separately; retries do not fix invalid credentials.

API overview↗ · Message structure