Zyte

Zyte stands as one of the oldest continuously running web scraping services, born from Scrapinghub's founding in 2010 in Cork, Ireland. The company rebranded under the Zyte name to reflect its evolution from a pure scraping tool into a broader data extraction platform. The service targets developers, data scientists, and enterprises who need production-grade infrastructure to collect data from defended websites at scale, handling the proxy rotation, JavaScript rendering, and anti-bot evasion themselves rather than forcing users to build that stack from scratch.

The platform divides its offering into two main request types: standard HTTP requests and browser-based requests that execute JavaScript. Both models support the core use case of sending a URL and receiving rendered HTML or structured data back, along with optional features like automatic visual element extraction, network behavior capture, and screenshots. Zyte promises deterministic retries on failures and handles session management across requests (important for workflows that need to log into sites or maintain consistent browser fingerprints across multiple related requests). The service also manages the tier assignment of each target domain automatically, reviewing classifications quarterly with two weeks notice to users before any tier change takes effect.

Pricing is where Zyte becomes notably complex, and this is the primary reason for its B-minus grade. The service operates on a tiered model where the cost of a single request depends on the target website, not just the request type. Zyte assigns each domain to one of five tiers for HTTP requests and five separate tiers for browser requests, with lower tiers costing less per request. The company does not publish the per-request costs for each tier or disclose which domains fall into which tiers upfront; instead, users discover pricing through trial, dashboard inspection, or the API. This opacity creates surprises during the scaling phase, when previously affordable sites suddenly cost substantially more. Monthly plans range from a $100 base with a 3,000-request-per-minute cap, up to $2,500 commitments with a 10,000 RPM cap, and all plans include discounts for annual prepayment. The critical point is that you are charged only for successful responses; rate-limiting, authentication failures, and malformed requests incur no cost, which is fair but does not offset the difficulty of predicting actual spend when tiering is hidden.

The service offers additional modular features that layer costs on top of the base request charge. Network captures, screenshots, and arbitrary action execution each cost $0.002 per request. Automatic data extraction costs between $0.0004 and $0.0016 per attribute extracted, depending on the extraction method. Custom attributes built on top of the standard model carry variable costs. These additions are genuinely useful (the extraction layer can save weeks of post-processing work), but each one is a cost decision the user must make blindly or through careful budget testing. Volume discounts exist and reach as high as 52 percent for $500 monthly commitments, but negotiating them requires contact with sales, and the discount is applied to the commitment amount, not necessarily to your actual usage.

On a practical level, Zyte's documentation is thorough and the API is straightforward to call from any language with HTTP support. The SDK for Python and Node is mature and widely used, so a developer can get running quickly once they understand the pricing model. The success rate reported in independent testing places Zyte at 93.14 percent across 15 heavily protected sites tested at production concurrency, which is respectable but notably lower than competing services like Scrapfly, which achieves 98 percent on the same tests. For most websites and e-commerce sites running standard Cloudflare or Akamai protection, Zyte succeeds most of the time, but high-difficulty targets like financial services sites with custom anti-bot logic or sites running multiple overlapping detection systems see substantially more failures.

The company has built a reputation for strong documentation and for maintaining backward compatibility with legacy API versions, which matters for teams with established code. However, user feedback across forums and independent reviews notes slower support response times at lower pricing tiers and occasional delays in addressing blocking issues. Some users report that support requests can take days to resolve, and escalation requires either a premium support add-on or a sufficiently large monthly commitment to get priority access. For startups or individual developers testing the service on a tight budget, this slow support can become frustrating when a legitimate blocking issue appears.

Zyte's platform also offers data extraction APIs and a cloud browser service that operate independently of the core scraping API, which can be useful for specific workflows. The extraction service trains models on your data samples, improving accuracy over time, but this requires feedback loops and ongoing tuning that smaller teams often find time-consuming. The cloud browser is a thin layer on top of standard browser automation libraries and is primarily useful for edge cases where you need persistent browser state across many requests; for most use cases, the standard rendering API is sufficient.

The right use case for Zyte is a team with medium to large budgets, a predictable set of target sites, and enough engineering resources to negotiate tiering and monitor spend carefully. The platform excels at handling high volumes of requests across many websites in a single organization, and the volume discounts make sense at scale. For teams running thousands of daily requests against a stable set of domains, Zyte's cost predictability improves over time as you learn the tiers. If your target sites are mostly straightforward (retail, news, publishing) without advanced anti-bot systems, Zyte is a reliable option.

The B-minus grade reflects the trade-offs Zyte makes: it is a serious, production-tested platform with a strong track record, but the hidden pricing tiers, opaque per-request costs, slower support at lower tiers, and slightly lower success rates on defended sites compared to newer competitors like Scrapfly create friction that many teams find irksome. For teams that have already chosen Zyte or have existing integrations deeply embedded in their code, the platform works reliably. For new projects, however, alternatives with transparent pricing and higher anti-bot success rates may be worth the evaluation cost.