Integration guide
Anthropic Python SDK and Claude Integration Guide
Call Claude with the Anthropic Python SDK: connect to ZHUAN's /proxy endpoint, use your ZHUAN API key and send requests in Messages format.
Installation and endpoint
Prepare an effective ZHUAN key and set ZHUAN_API_KEY. Use https://key.zhuan.xyz/proxy as the SDK base_url. The SDK appends /v1/messages, so do not repeat the full request path in base_url.
python -m pip install anthropic
A minimal Messages request
max_tokens limits the output for this request, while messages contains the user input. This example uses claude-sonnet-4-6. Confirm it is still in the catalog and check current prices first.
import os
from anthropic import Anthropic
client = Anthropic(
api_key=os.environ["ZHUAN_API_KEY"],
base_url="https://key.zhuan.xyz/proxy",
timeout=60.0,
max_retries=0,
)
message = client.messages.create(
model="claude-sonnet-4-6",
max_tokens=128,
messages=[{"role": "user", "content": "Hello"}],
)
for block in message.content:
if block.type == "text":
print(block.text)
Streaming Messages
Reuse the client above and read text_stream inside the messages.stream context manager. The example handles text; tool calls require handling structured content blocks separately.
with client.messages.stream(
model="claude-sonnet-4-6",
max_tokens=128,
messages=[{"role": "user", "content": "Hello"}],
) as stream:
for text in stream.text_stream:
print(text, end="", flush=True)
Compatibility and troubleshooting
- Validate system prompts, caching, tools and multimodal parameters against the current API. Workloads requiring strict instruction following should be checked with your own evaluation set.
- For 401, check your ZHUAN key and x-api-key header. For 429, reduce concurrency and back off according to the response. Do not mix OpenAI and Messages request fields.
- After a successful call, check the model, tokens and charge in the console. Retain the request ID for debugging, not the complete key.