ProxyScrape
- Free allowance
- Free, no account: the page states 20,000+ live proxies.
- Verdict
- deadthe offer or the service no longer exists
- Cost
- no price stated
- Identity
- anonymousno card, no phone
- Binding limit
- request-rate2,000 rows per response on format=text, against the page's claim of 20,000+ live proxies: the body was exactly 2,000 lines
- Evidence
- measured4 sources; a live request was made and its output kept
- Freshness
- checked 2026-10-24checked 2026-10-24 (live)
| Free tier | Free, no account: the page states 20,000+ live proxies. The format=text endpoint returned exactly 2,000 rows on 2026-10-24, so the offer a client can actually pull is 2,000 addresses per response: 1,357 SOCKS5, 409 HTTP, 234 SOCKS4. unit requests; metered on none; The 2,000 is a per-response row cap, not a metered allowance: nothing is charged and nothing is counted against a user. Two pulls five minutes apart were both served, with no refusal observed in either. |
|---|---|
| Price | none stated |
| Limits |
Two further constraints were found that name no vocabulary kind, so they are kept here rather than forced into one. First, yield: of 100 sampled addresses, 38 completed a handshake and 32 carried a payload, and five minutes later 34 and 30, so about 70% of the list cannot move a byte at the moment it is pulled. Second, attribution: the response carries no ASN field, so there is no way to know whose network an address belongs to without spending a request on it. That absence is what puts rung 5 out of reach before the first use. |
| Binding | request-rate The response is capped at 2,000 rows against an advertised 20,000+, so the ceiling a client can reach is the size of one response, not the size of the published pool; that cap arrives before any byte quota, which does not exist, and before any per-address shaping, which no evidence here shows. The 30% payload rate then cuts the working set to roughly 600 addresses, which is the figure that actually decides the gateway, but it is a yield property of the addresses rather than a limit the vendor sets. |
| Identity | anonymous No account, key or email: the format=text endpoint answered an unauthenticated GET on 2026-10-24 and the 2,000-row body was used as returned. |
| Egress | ProxyScrape prices nothing and states nothing about traffic: it publishes the list, and the traffic itself crosses the third-party address, whose operator's terms are not published anywhere. free allowance: none published |
| Retention | not applicable: a live proxy address, not stored bytes |
| Interface | rest One unauthenticated HTTP GET on api.proxyscrape.com/v4/free-proxy-list/get, which returned Content-Type text/plain with Transfer-Encoding chunked. The addresses it returns speak socks5, http and socks4 (1,357 / 409 / 234 of the 2,000 rows). The schema's interface enum has no token for a proxy protocol, so the three protocols are named here rather than in the array. endpoint: https://api.proxyscrape.com/v4/free-proxy-list/get (format=text) |
| Tooling | A single curl, or any HTTP client, pulls the list; no account, key, SDK or wrapper is involved. Using an address afterwards needs a client that speaks the protocol it is published under (SOCKS5, SOCKS4, or HTTP CONNECT), which curl and rclone both do natively. |
| Composes with | It is a supply of last-resort egress addresses for the Workers AI gateway: a request a Worker cannot make from its own address is retried through a published address, at roughly a 30% payload rate and with an exit network nobody has identified. |
| Reliability | The addresses are open proxies of unknown ownership: 32 of 100 sampled carried a payload and 30 of 100 five minutes later, the response publishes no ASN field, and the exit network is therefore unknown until an address is actually used. |
| Notes | The verdict rests on (2) with (3) still open. One caveat on weight: each rate is one sample of 100 addresses out of 2,000, taken twice five minutes apart, which is enough to show the yield is low and repeatable and not enough to put a tight interval on it. The 2,000-row cap rests on a single measured response whose length is exactly the round number, which is what a server-side cap looks like; Rung 5 is unknowable before a request is spent, because the response has no ASN field; that absence is a property of the product, not a gap in the probe. None exists, so this record corrects nothing outside itself. |
| Evidence note | measured The 2,000-row cap, the protocol split and the two payload rates were each measured from the endpoint's own output on 2026-10-24, and the measurement is what contradicts the page's 20,000+ claim. Moving up one would need a sample larger than 100 and runs spread across a day, so that a diurnal effect can be told apart from a dead list. |
| Parsed numbers |
|
Evidence 4
| date | source | what it proved |
|---|---|---|
| 2026-10-24 | proxyscrape.com/free-proxy-list | The vendor's own figure for the free list is 20,000+ live proxies, which the measured 2,000-row format=text response contradicts. |
| 2026-10-24 | api.proxyscrape.com/v4/free-proxy-list/get | format=text answered an unauthenticated GET with exactly 2,000 lines: 1,357 socks5, 409 http, 234 socks4, served as text/plain with Transfer-Encoding chunked. No ASN field appears in the response. |
| 2026-10-24 | a probe; its output is not published | First run over 100 sampled addresses: 38 completed a handshake and 32 carried a payload, 32%, with the failures logged as connection resets, EOFs and wrong-version TLS errors. |
| 2026-10-24 | a probe; its output is not published | Second run five minutes later over 100 sampled addresses: 34 handshakes and 30 payloads, 30%, so the low yield repeats rather than being a single bad sample. |
More fields 11
| rung | 2 |
|---|---|
| rung note | The product is a supply of single addresses rather than a subnet or an ASN block, so rung 2 is the furthest a user can get before spending a request on an address. |
| protocol split | {"socks5": 1357,"http": 409,"socks4": 234,"total": 2000} |
| payload rate run1 | 32 of 100 |
| payload rate run2 | 30 of 100 |
| handshake rate run1 | 38 of 100 |
| handshake rate run2 | 34 of 100 |
| asn field | no |
| advertised count | 20,000+ live proxies |
| served count | 2000 |
| sample size | 100 addresses per run, two runs five minutes apart |
Unknown values 1
- per object cap stated: the field does not apply to this kind of offer