Proxy list API
Free and keyless, no signup. Base URL https://api.proxyscraper.com. Every response says which snapshot it comes from; a new one arrives every 10 minutes. More than 10 requests in 10 seconds, or every country at once? A free key.
Start here
Fetch a list, try its first five proxies, and print the address a site sees through the first that works:
(for P in $(curl -fsS "https://api.proxyscraper.com/v1/socks5.txt" | head -n 5); do
echo "trying $P" >&2
curl -sS --max-time 10 -x "socks5h://$P" "https://api.proxyscraper.com/v1/ip" && exit 0
done; exit 1)Only proxies whose exit is in one country, faster than 1 second:
curl -s "https://api.proxyscraper.com/v1/http/id.txt?max_latency=1000"Leave out hosting and cloud exits, and keep proxies that were up at least half of the last day:
curl -s "https://api.proxyscraper.com/v1/socks5.txt?network=other&min_uptime=50"The list page builds these URLs as you filter.
Endpoints
| Path | Returns | Try it |
|---|---|---|
/v1/{socks5|http}.txt | One ip:port per line, fastest first, at most one per exit address. ?format=uri gives socks5://ip:port. | socks5.txt (47 now) |
/v1/{socks5|http}.json | The same rows with every field (below), in an envelope with the snapshot time. | http.json (248 now) |
/v1/{socks5|http}/{cc}.txt, .json | Only rows whose exit is in that country: ISO 3166-1 alpha-2, lower case (us, de, gb). A valid code without rows gives an empty list. Every country in one request: countries.json under a key. | socks5/id.txt (3 now) |
/v1/{socks5|http}/stable.txt, .json | Rows up in at least 90% of our snapshots over 7 days. | socks5/stable.txt (2 now) |
/v1/ip | The IP address you call from, as plain text. Use it to see what a site sees through a proxy. | ip |
/v1/health | Snapshot time, age and counts as JSON; 503 while the data is paused. | health |
Parameters
| Name | Values | Applies to |
|---|---|---|
max_latency | 500, 1000, 1500, 5000 (ms) | Protocol and country lists. Not stable lists. |
network | hosting: exits on hosting or cloud networks. other: every other exit, unknown networks included (the label) | Protocol and country lists. Not stable lists. |
min_uptime | 50, 90: passed in at least that share of our snapshots over the last 24 hours. New proxies without a 24-hour figure are left out | Protocol and country lists. Not stable lists. |
format | ipport (default) or uri | The .txt lists. |
Filters combine: https://api.proxyscraper.com/v1/socks5/us.txt?max_latency=1500&network=other&min_uptime=50. Any other parameter is ignored.
For proxies that tend to last, add min_uptime=50: in our lists a proxy's 24-hour uptime predicts whether it is still listed ten minutes later far better than its latency does.
JSON fields
The envelope is {snapshot, generation, count, proxies}. Each proxy has:
| Field | Type | Meaning |
|---|---|---|
ip, port | string, integer | The address you connect to. |
protocol | string | socks5 or http. HTTP proxies here support CONNECT, so they work with https:// sites. |
latency_ms | integer | Time for our HTTPS check through the proxy, measured from Los Angeles. |
last_checked | string (UTC) | When the latest check passed: at most 15 minutes before the snapshot. |
uptime_24h, uptime_7d | number or null | Share of our snapshots, since the proxy first appeared, in which it passed. null until 6 (24 h) or 144 (7 d) snapshots exist. |
first_seen | string (UTC) | The first snapshot that listed it. |
country | string or null | DB-IP Lite's country for exit_ip, upper case. |
exit_ip | string or null | The address our server saw the check arrive from: what sites see. It differs from ip for relays. |
asn, asn_name | integer, string or null | The autonomous system of exit_ip and its registered name (DB-IP ASN Lite). |
network_type | string or null | "hosting" when that AS is a hosting or cloud provider (our list and rule); null otherwise, which does not mean residential. More values may come; treat unknown ones as null. |
Headers
| Header | Meaning |
|---|---|
X-Snapshot-Time | The snapshot the body belongs to (UTC). |
X-Proxy-Count | Rows in the body. |
ETag, Last-Modified | Send If-None-Match or If-Modified-Since: 304 until the next snapshot. |
Cache-Control | public, max-age=60. |
RateLimit-Policy | 60;w=60: our server's per-minute limit. Two more limits apply; see 429 below. Under a key: 1000;w=60, 100000;w=86400. |
Access-Control-Allow-Origin | *: browsers may fetch the lists from any site. |
Errors
Errors are plain text, one line, except one kind of 429.
| Status | When | Body |
|---|---|---|
| 400 | An unknown value for max_latency, network, min_uptime or format; an unknown or upper-case country code; a filter a stable list doesn't take. | One line saying what to change. |
| 404 | A path that isn't listed here. | The valid shapes. |
| 429 | Too many requests from one IP address. Cloudflare's edge allows no more than 10 requests to /v1/ URLs in 10 seconds, cached answers included. Requests that reach our server (both hosts) are refused at more than 5 a second after a burst of 30, and at more than 60 a minute. One pull of every protocol and stable list, both formats, plus /v1/ip and /v1/health is 10 requests; poll once per snapshot. Need more? A free key. | Edge: error code: 1015 with Retry-After: 10. Per second: one line naming the key page, with Retry-After: 10 (under a key, an HTML error page). Per minute: the rate limit line with Retry-After: 60, naming the key page. |
| 503 | The newest snapshot is older than 2 hours, or none exists yet. | Retry-After: 600. Old lists are never served. |
More than 10 requests in 10 seconds? A free key
Every /v1/ URL also works under a key, with the same parameters and the same answer: https://api.proxyscraper.com/k/{key}/v1/…. Sign in with Google on the key page to get one. The free URLs stay as they are.
| Free, no key | With a free key | |
|---|---|---|
| URLs | https://api.proxyscraper.com/v1/… | https://api.proxyscraper.com/k/{key}/v1/…: every /v1/ path and parameter, the same answer |
| Cloudflare's limit | 10 requests in 10 seconds per IP address | Does not apply |
| Our server's limits | 5 a second after a burst of 30, and 60 a minute, per IP address | 20 a second after a burst of 100, per key and per IP address; 1,000 a minute and 100,000 a day per key |
| Country lists | One country per request | Every country in one document: /k/{key}/v1/{protocol}/countries.json |
| Earlier snapshots | The current one only | The ones we keep, about the last two hours: /k/{key}/v1/snapshots/{generation}/… |
| History | uptime_24h and uptime_7d per row | Which of the last 7 days' snapshots listed each row: /k/{key}/v1/{protocol}/history.json |
| Rows and freshness | As published | The same: a key never gets fresher, more or better rows |
Keep the key in an environment variable rather than in code you publish. Every protocol list, then every country at once:
curl -s "https://api.proxyscraper.com/k/$PROXY_SCRAPER_KEY/v1/socks5.txt"curl -s "https://api.proxyscraper.com/k/$PROXY_SCRAPER_KEY/v1/socks5/countries.json"The command-line scraper (v0.2.0 and later) reads the same variable: with it set, proxy-scraper fetch asks the keyed URLs, at the key's limits. -key psk_… overrides it for one run, but a flag shows in ps and shell history. Only fetch sends the key.
An unknown or revoked key answers 404; over a cap, 429 with Retry-After.
Versioning and changes
Published /v1/ URLs never change meaning, and fields are only ever added. A breaking change would ship as /v2/ with /v1/ kept alive.
- 2026-10-21: the parameters
networkandmin_uptimeon protocol and country lists. - 2026-10: country lists (
/v1/{protocol}/{cc}.txt); the fieldsasn,asn_nameandnetwork_type. - 2026-10-16: v1 launched: protocol and stable lists,
/v1/ip,/v1/health.