Calling an LLM API

A real call

With an SDK, a call is a few lines. This is what it looks like with the Anthropic Python SDK (pip install anthropic, key in the ANTHROPIC_API_KEY environment variable):

import anthropic

client = anthropic.Anthropic()                       # reads ANTHROPIC_API_KEY
response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=300,
    system="You are a concise assistant.",
    messages=[{"role": "user", "content": "Explain HTTP caching in two sentences."}],
)
print(response.content[0].text)
print(response.usage.input_tokens, response.usage.output_tokens, response.stop_reason)

The response

• content: a list of blocks; plain answers have one text block.
• usage: input and output tokens, for cost tracking.
• stop_reason: why generation stopped: end_turn (finished), max_tokens (cut off), tool_use (wants a tool), and others.

responses = [
    {"text": "HTTP caching stores responses...", "stop_reason": "end_turn", "usage": (42, 38)},
    {"text": "HTTP caching stores responses so that repeated requests can be served without", "stop_reason": "max_tokens", "usage": (42, 20)},
]
for r in responses:
    note = "complete" if r["stop_reason"] == "end_turn" else "CUT OFF: raise max_tokens or ask for less"
    print(f"{r['stop_reason']:10} in={r['usage'][0]} out={r['usage'][1]}  {note}")

end_turn   in=42 out=38  complete
max_tokens in=42 out=20  CUT OFF: raise max_tokens or ask for less

Keep keys safe

API keys belong in environment variables or a secrets manager on the server. Never put them in browser or mobile code, and never commit them. See the API overview (Anthropic's docs; other providers work the same way).