ScrapingFish

ScrapingFish is a web scraping API rather than a raw proxy network, built around a single promise that is unusual in this market: you pay $0.002 per successful request, and every request runs through a rotating 4G/LTE mobile IP with JavaScript rendering included by default. It is built by Mateusz Buda and Pawel Kobojek under the Narf AI brand and targets developers and small-to-mid scraping teams who want a predictable per-request bill instead of credit math, premium-proxy multipliers, or per-GB residential pricing.

Under the hood, ScrapingFish does not sell access to a proxy pool you control. Instead, every call is routed through their managed cluster of real browsers backed by rotating mobile proxies sourced, at the time of writing, from Polish 4G/LTE carriers. The official site explicitly notes the company is working on global geolocation but is not there yet, so if you need US, UK, Germany, or APAC exit nodes, this is the wrong tool. There is no published pool size, no city-level targeting, and no SOCKS5 endpoint to plug into a proxy manager — the only interface is the HTTP API.

Pricing is the strongest part of the pitch. Standard packs are $2 for 1,250 requests, $20 for 10,250 requests, $200 for 105,000 requests, and $2,000 for 1.1M requests, with expiration windows from one month to one year. There is also a flat $100/month unlimited subscription per scraping concurrency, which is the cheapest way to run heavy parallel jobs if your target sites are not heavily protected. Payment runs through Stripe. There is no free trial in the traditional sense, but you can buy the smallest $2 pack to test end-to-end with no account commitment, and the bonus credits on every tier function as a soft trial.

Feature-wise, ScrapingFish leans simplicity over surface area. You get JavaScript rendering with no premium multiplier, a JavaScript scenarios feature for clicks, typing, and custom in-page code, and a documented Python client. The G2 and SaaSHub reviews consistently call out responsive support and a published 99.96% success rate benchmark on common e-commerce targets. What you do not get: built-in CAPTCHA solving, anti-bot fingerprint stacks tuned for Cloudflare or DataDome, geographic targeting, sticky session controls, or a dashboard with the analytics depth of Bright Data or Oxylabs. It is a thin, fast, cheap API — not a platform.

The ideal ScrapingFish customer is a Python or Node developer scraping product catalogs, marketplaces, public listings, SERPs (non-Google), or social profiles at 10K to 1M requests per month, who is tired of ScrapingBee or ScraperAPI charging 5x credits for JS rendering and stealth mode. At 100K requests a month you pay $200 here versus roughly $1,500 on stealth tiers elsewhere — that gap is real. Skip ScrapingFish if you need a true residential proxy network you control, geo-targeted IPs outside Poland, sneaker-grade rotating mobile pools, or aggressive anti-bot bypass for Cloudflare-protected targets. For those, Decodo, Bright Data, Oxylabs, or NetNut are better fits.

Limitations worth knowing up front: the Poland-only egress is a hard ceiling for any geo-sensitive work, the dashboard is minimal, and there is no formal SLA published. Pricing also assumes your targets cooperate — if a site fights back with serious anti-bot, the success rate drops, and because billing is per successful request the math still works for you, but throughput will not. Verify your specific targets with the $2 starter pack before committing to a 1M-request annual pack.