Stream Claude Sonnet 5.5 replies from a FastAPI endpoint

What you get

A FastAPI endpoint that streams Claude Sonnet 5.5 replies as plain text using the official Anthropic Python SDK.

Prerequisites

A Claude API key in ANTHROPIC_API_KEY (the SDK reads it automatically)
Python 3.10 or later with anthropic, fastapi and uvicorn

Steps

Install packages and set your key.

Shell


pip install anthropic fastapi uvicorn

export ANTHROPIC_API_KEY="your-api-key-here"

Create main.py. client.messages.stream(...) yields text chunks via text_stream.

Python


import anthropic

from fastapi import FastAPI

from fastapi.responses import StreamingResponse

from pydantic import BaseModel



app = FastAPI()

client = anthropic.AsyncAnthropic()



class ChatIn(BaseModel):

    prompt: str



@app.post("/chat")

async def chat(body: ChatIn):

    async def gen():

        async with client.messages.stream(

            max_tokens=1024,

            messages=[{"role": "user", "content": body.prompt}],

            model="claude-sonnet-5-5",

        ) as stream:

            async for text in stream.text_stream:

                yield text



    return StreamingResponse(gen(), media_type="text/plain")

Run the server.

Shell


uvicorn main:app --reload

Test it

Shell


curl -N -X POST http://localhost:8000/chat \

  -H "Content-Type: application/json" \

  -d '{"prompt":"Hello"}'

You should see Claude's reply arrive incrementally.