How to build a chatbot with Claude API in python
Direct answer
Use the
anthropic Python SDK to create a client with your API key, then call client.messages.create with the claude-3-5-sonnet-20241022 model and a messages array to build a chatbot.Setup
Install
pip install anthropic Env vars
ANTHROPIC_API_KEY Imports
import os
import anthropicExamples
inHello, who won the 2024 US presidential election?
outThe 2024 US presidential election was won by Joe Biden.
inCan you help me write a Python function to reverse a string?
outSure! Here's a Python function to reverse a string:
```python
def reverse_string(s):
return s[::-1]
```
inTell me a joke about computers.
outWhy do programmers prefer dark mode? Because light attracts bugs!
Integration steps
- Install the anthropic Python SDK and set your API key in the environment variable ANTHROPIC_API_KEY.
- Import the anthropic library and initialize the client with your API key from os.environ.
- Create a messages list with user input as a dictionary with role 'user' and content as the message text.
- Call client.messages.create with the model 'claude-3-5-sonnet-20241022', system prompt, max_tokens, and the messages list.
- Extract the chatbot's reply from the response's content field and display it.
Full code
import os
import anthropic
client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])
system_prompt = "You are a helpful assistant chatbot."
messages = [
{"role": "user", "content": "Hello! Can you tell me a fun fact about space?"}
]
response = client.messages.create(
model="claude-3-5-sonnet-20241022",
max_tokens=300,
system=system_prompt,
messages=messages
)
print("Chatbot reply:")
print(response.content[0].text) output
Chatbot reply: Sure! Did you know that a day on Venus is longer than a year on Venus? It takes Venus about 243 Earth days to rotate once, but only about 225 Earth days to orbit the Sun.
API trace
Request
{"model": "claude-3-5-sonnet-20241022", "max_tokens": 300, "system": "You are a helpful assistant chatbot.", "messages": [{"role": "user", "content": "Hello! Can you tell me a fun fact about space?"}]} Response
{"id": "chatcmpl-xxx", "object": "chat.completion", "created": 1680000000, "model": "claude-3-5-sonnet-20241022", "choices": [{"index": 0, "message": {"role": "assistant", "content": ["Sure! Did you know that a day on Venus is longer than a year on Venus? It takes Venus about 243 Earth days to rotate once, but only about 225 Earth days to orbit the Sun."]}, "finish_reason": "stop"}], "usage": {"prompt_tokens": 20, "completion_tokens": 50, "total_tokens": 70}}Extract
response.content[0].textVariants
Streaming Chatbot with Claude API ›
Use streaming to display chatbot responses token-by-token for better user experience in interactive applications.
import os
import anthropic
client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])
system_prompt = "You are a helpful assistant chatbot."
messages = [{"role": "user", "content": "Tell me a story about a robot."}]
for chunk in client.messages.stream(
model="claude-3-5-sonnet-20241022",
max_tokens=300,
system=system_prompt,
messages=messages
):
print(chunk.content[0].text, end='', flush=True)
print()Async Chatbot with Claude API ›
Use async calls when integrating the chatbot in asynchronous Python applications or web servers.
import os
import asyncio
import anthropic
async def main():
client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])
system_prompt = "You are a helpful assistant chatbot."
messages = [{"role": "user", "content": "Explain quantum computing in simple terms."}]
response = await client.messages.acreate(
model="claude-3-5-sonnet-20241022",
max_tokens=300,
system=system_prompt,
messages=messages
)
print("Chatbot reply:", response.content[0].text)
asyncio.run(main())Use Claude 3 Opus Model for Cost Efficiency ›
Use the Claude 3 Opus model for faster responses and lower cost when ultra-high accuracy is not critical.
import os
import anthropic
client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])
system_prompt = "You are a helpful assistant chatbot."
messages = [{"role": "user", "content": "Summarize the latest tech news."}]
response = client.messages.create(
model="claude-3-opus-20240229",
max_tokens=300,
system=system_prompt,
messages=messages
)
print("Chatbot reply:", response.content[0].text)Performance
Latency~800ms for typical 300-token response on claude-3-5-sonnet-20241022
Cost~$0.003 per 300 tokens on claude-3-5-sonnet-20241022
Rate limitsTier 1: 300 requests per minute / 20,000 tokens per minute
- Keep system prompts concise to reduce token usage.
- Limit max_tokens to control response length and cost.
- Reuse conversation context selectively to avoid token bloat.
| Approach | Latency | Cost/call | Best for |
|---|---|---|---|
| Standard Chatbot (claude-3-5-sonnet-20241022) | ~800ms | ~$0.003 | High-quality, general chatbot |
| Streaming Chatbot | ~800ms initial + streaming | ~$0.003 | Interactive apps needing token-by-token output |
| Async Chatbot | ~800ms async | ~$0.003 | Concurrent or async Python apps |
| Claude 3 Opus Model | ~600ms | ~$0.0015 | Cost-sensitive or faster responses |
Quick tip
Always provide a clear system prompt with <code>system=</code> to guide Claude's chatbot behavior effectively.
Common mistake
Beginners often forget to use the <code>system=</code> parameter and instead try to include system instructions as a message, which Claude API does not support.