The whole live list in one call
One GET returns every working proxy we have, as txt, json or csv. Not a sample and not a teaser: HTTP, HTTPS, SOCKS4 and SOCKS5, all of it, in a single call. No key, no signup. CORS is enabled, so you can call it straight from a browser. Fair-use rate limit of about 120 requests per minute per IP, with bursts up to 120 (a clean 429 with Retry-After beyond that). This is the same data behind the free proxy list.
How many is that? Thousands at any moment, and tens of thousands within two days, because that is how many public proxies on Earth are actually answering. It moves every few minutes as proxies die and new ones are found, so treat any number written here as a rough order of magnitude and never as the truth. X-Total-Available on the response is the exact count matching your filter, and X-Total-Count is how many came back in this particular response. If the two are equal you already have everything and there is no next page.
Why you will see 10,000 mentioned below. That is the most rows a single response will carry when you ask for a specific number. It is a page size, not a cap on what you are allowed to have. Omit limit entirely and you get the complete matching set in one response, however large it is. The page size only comes into play when you pass limit yourself, or on a set big enough that we page it for you (recent=true).
Request
- Base URL: https://hproxy.com
- URL: https://hproxy.com/api/proxy-list
- Auth: none required
Query parameters
Response shape. Default txt (one ip:port per line).
Filter by protocol. Comma-separated allowed (e.g. http,https), which matches a proxy supporting ANY of them.
Filter by country, e.g. US, DE, GB.
Filter by anonymity grade. Proxies we have not been able to grade yet are excluded.
Filter by autonomous system, e.g. 14061 for DigitalOcean. Same data as the per-network pages on the free list.
Only proxies answering on this exact port, e.g. 8080 or 1080.
Only proxies inside one /24 block, e.g. subnet=45.61.188 for 45.61.188.0/24. Open proxies cluster in blocks, so this finds a machine's neighbours.
Only proxies whose lifetime uptime is at least this. The single most useful filter if you want proxies that stay up.
Only proxies at or below this latency, in milliseconds.
Order the whole pool by reliability before paging, so page 1 is the best proxies overall. Omit for the default order (alive first, then fastest).
Rows in THIS response, not a limit on your total. Leave it out and you receive the complete matching set in one call, which is what most callers want. Pass a number and this response stops at that many rows (10000 is the ceiling for one response); use offset to walk the rest. Whatever you do, X-Total-Available still reports how many exist.
Skip this many rows. Needed only when you passed an explicit limit, or on a set too large for one response. Default 0.
Also include proxies that have gone quiet but answered within the last 48h, taking the set from the live-now pool to several times its size. This WIDENS the list, it does not freshen it: the extra rows come back with status recently_alive and many are not answering right now. Leave it off for the strictly alive-now pool.
Response
Success
Changes
No recorded changes to this operation across all 1 revision of this API.