Rotating residential
For workloads that need residential routing and session configuration. Estimate traffic from retries and browser assets alongside compute and maintenance costs.
Explore residential proxies →Compare scraping options
Choose between a managed scraping API, your own crawler with proxies, and interactive Browser/MCP tasks. Compare inputs, outputs, operating responsibility and costs.
Official sources checked: 2026-09-05
API, proxy or browser?
| Dimension | Crawler API | Proxy | Browser / MCP |
|---|---|---|---|
| Input | Interface key + parameters | Your HTTP requests + proxy credentials | AI instructions + browser actions |
| Output | Task status, result and credits | Proxied responses; your parser extracts the data | Browser observations and task-specific output |
| Responsibility | Service executes; you validate the result | You own scheduling, rendering, parsing and retries | AI client orchestrates tools; you handle verification |
| Best fit | Scheduled collection and content pipelines | Custom crawlers and existing infrastructure | Multi-step interaction, testing and human takeover |
| Cost unit | Credits by interface and parameters | Traffic or IP/time/bandwidth, by product | Current MIYAIP guidance: one call per successful task start |
| Connection | HTTPS + console login token | Proxy authentication in your own client | Streamable HTTP + dedicated MCP key |
A credit is not a useful result. Compare actual spend against data that passes your acceptance checks. Read the cost guide →
Proxies for your own crawler
A proxy supplies a network route. Your application still chooses URLs, runs a browser if needed, parses data, limits retries and stores results. This fits teams with an existing crawler or custom execution requirements.
Datacenter proxies
Consider this option when datacenter routing meets the target requirements. Compare IP and duration terms; routing alone does not guarantee target acceptance.
Explore datacenter proxies →Browser automation with MCP
Use an AI client to operate a remote browser for navigation, extraction, screenshots or testing. Choose this model when the task needs interaction rather than a single predefined data interface.
MIYAIP Browser / MCP BETA
Connect with a dedicated MCP key over Streamable HTTP. Current product guidance counts a successful browser_start_task or browser_run_test start as one call; actions within that task do not add calls. Each running task uses a concurrency slot.
Explore Browser/MCP →Timeouts and verification
Set an adequate tool timeout, validate the output and use human takeover for CAPTCHA, QR or SMS verification. A created key does not imply available task credits or concurrency.
Configure your MCP client →Compare API providers
Each guide covers who should switch, what it costs and what changes in your code.
ScraperAPI vs MiyaIP
Compare package costs and request formats for small projects.
ScrapingBee vs MiyaIP
Compare JavaScript rendering, credits and concurrency.
ZenRows vs MiyaIP
Check target-page support and costs under the same conditions.
Scrapfly vs MiyaIP
Review session handling, response parsing and migration changes.
Try a crawler task
Review the interface, choose a package and test the pages you actually need.