TopolearnHTTPX is a drop-in modern client; the basics look like Requests, so you are productive immediately.
Install with pip install httpx and import it as import httpx. The top-level functions mirror HTTP methods: httpx.get, post, put, patch, delete, head, options, plus the generic httpx.request(method, url). Each returns an httpx.Response.
The API was deliberately designed to feel like Requests, so most simple code translates directly:
r = httpx.get("https://httpbin.org/get", params={"q": "httpx"})
r.status_code, r.json()Under the hood every top-level call creates a temporary Client, sends one request and closes the connection. That is fine for scripts and one-off calls. For repeated calls you will use a Client (next module) to get connection pooling.
What HTTPX adds over Requests, as a preview:
- an async API (
httpx.AsyncClient) with the same methods, - timeouts by default (5 seconds),
- optional HTTP/2,
- transports that can route requests straight to a Python web app or a mock function, which makes testing easy,
- full type annotations.
Like Requests, an error status such as 404 does not raise automatically; you get a normal response and decide what to do.
The current release line is 0.28. HTTPX is still pre-1.0, so a few APIs changed between minor versions (for example proxies= became proxy= and app= was removed in 0.28); check the changelog when upgrading.
Example
import httpx
r = httpx.get("https://httpbin.org/get", params={"q": "httpx"})
print(r.status_code) # 200
print(r.json()["args"]) # {'q': 'httpx'}
print(r.http_version) # HTTP/1.1
r = httpx.post("https://httpbin.org/post", json={"lang": "python"})
print(r.json()["json"]) # {'lang': 'python'}
r = httpx.get("https://httpbin.org/status/404")
print(r.status_code, r.is_success) # 404 False (no exception)Remember
- httpx.get/post/... mirror Requests and return a Response
- Top-level calls open and close a client each time
- Error statuses don't raise automatically