The Thordata SERP API is a search engine scraping service that returns real-time results from Google, Bing, Yandex and DuckDuckGo as structured JSON or raw HTML, with proxies, CAPTCHAs and parsing handled entirely behind the scenes. It comes from Thordata, a Hong Kong-based proxy and web data platform that has been operating since 2018.
What makes the service interesting to developers is its billing rule: only successful HTTP 200 responses are charged. Blocked requests, timeouts and error codes return nothing to your bill, which flips the usual economics of mass SERP collection on their head. A seven-day free trial ships 500 responses and asks for no credit card.
This guide walks through what the Thordata SERP API actually covers, how requests and geotargeting work, what the pricing tiers cost per thousand responses, how it compares with SerpApi, DataForSEO and Bright Data, and where its AI-native integrations, including an MCP server, fit modern agent pipelines. Numbers come from Thordata’s official product pages and documentation as of late October 2026.
Table of contents
- What Is the Thordata SERP API?
- How the Thordata SERP API Works
- From Sign-Up to First JSON Response
- Thordata SERP API Pricing: Only Success Is Billed
- Thordata SERP API vs SerpApi, DataForSEO and Bright Data
- Feeding AI Agents With the Thordata SERP API
- Limitations, Compliance, and What to Watch
- Should You Use the Thordata SERP API?
- FAQs About the Thordata SERP API
- References
What Is the Thordata SERP API?
The Short Answer
The Thordata SERP API takes a search query and returns the results page of a major search engine as machine-readable data. You send an engine, a keyword, and targeting parameters; the service handles the IP rotation, the anti-bot friction, and the parsing, and hands you structured results you can pipe straight into rank trackers, dashboards, or model-training datasets.
Unlike a bare proxy, it is a finished data product. The output arrives as JSON with fields already extracted, or as Light JSON if you only want the top organic results, or as raw HTML when your own parser needs the original markup. The scraping infrastructure, which Thordata also sells separately as residential, ISP, and mobile proxies, sits underneath every request.
Four Engines and Seventeen Google Verticals
One REST endpoint covers Google, Bing, Yandex and DuckDuckGo. Google alone exposes seventeen documented verticals: Search, Shopping, Maps, Hotels, Images, Trends, Local, News, Flights, Scholar, Videos, Jobs, Lens, Finance, Patents and Play. Bing adds six verticals of its own, covering Search, News, Shopping, Maps, Images and Videos.
That vertical coverage is the first place the Thordata SERP API separates from bargain-bin scrapers. Rank tracking is one job; pulling Google Maps listings, hotel availability, job postings, or Google Scholar entries is a different product surface entirely, and here it arrives through the same call pattern and the same billing meter.
Who Builds It, and Why That Matters
Thordata operates from Hong Kong and has run proxy infrastructure since 2018, advertising over 100 million residential IP addresses across 190-plus countries. The SERP API reuses that network, which is why the vendor can claim sub-second average response times and 99.9% uptime rather than promising infra it has to build from scratch.
For buyers, an established proxy operator is a meaningful signal. SERP scraping lives or dies on IP quality and block rates, and a provider with a decade of network operations behind it is better placed to keep both numbers healthy than a thin wrapper over someone else’s pool. You can track adjacent releases on our AI Models, Tools & Releases hub.
How the Thordata SERP API Works
One Request, One Structured Answer
You send a POST request carrying the search URL you want fetched, along with your token. Thordata queues the job on its network, renders the result page, extracts the fields for the selected vertical, and returns JSON. Average response time is quoted under one second, and the Playground shows the structured payload live so you can see the shape of the data before writing a line of integration code.
Concurrency is where the architecture shows its proxy heritage. Requests spread across the residential pool rather than hammering search engines from fixed IPs, which is what keeps block rates low enough for the pay-on-success model to make business sense. Failed deliveries cost Thordata, not you.
Geotargeting Down to the City
Every request accepts country, language, and city-level targeting through gl, hl, and location or uule parameters, across a claimed 195-plus countries and cities. That granularity matters because search results are local by design: a SERP for “plumber near me” in Berlin and the same query in Brisbane share nothing but the words.
Local SEO agencies are the obvious users, pulling Maps packs and local rankings city by city. E-commerce teams use the same mechanism to check Shopping carousels per market. The Thordata SERP API treats geography as a first-class parameter rather than an afterthought, which is exactly what any serious rank-monitoring workload demands.
Three Output Formats for Three Jobs
Full JSON is the default: every extracted field of the selected vertical, ready for storage or transformation. Light JSON trims the payload to the top organic results, which cuts bandwidth and parsing time when you only need rankings. Raw HTML hands you the untouched page when your in-house parser or QA process needs to see what a human would see.
Most teams pick one format per workload. Rank tracking runs on Light JSON, feature analysis runs on full JSON, and compliance archiving runs on raw HTML. Switching is a parameter change, not a re-integration, so the Thordata SERP API covers the whole ladder without three separate contracts.
What the Thordata SERP API Can Track Beyond Rankings
Once the call pattern exists, the verticals invite use cases your team probably runs manually today. Monitor a competitor’s Google Shopping listings per market, watch hotel availability for travel pricing, or pull job postings to time hiring waves. Google Trends through the same token gives search-demand context to every rank movement you log.
Scholar and Patents verticals serve research teams tracking citations and filings, and Lens coverage supports retail teams checking visual search results for product shots. Each is a separate meter on the same Thordata SERP API balance, so extending from rank tracking into adjacent intelligence costs integration effort, not a new vendor contract.
From Sign-Up to First JSON Response
The Playground Pulls Your First SERP
Getting started genuinely takes minutes. You register on the Thordata dashboard with an email, Google, or GitHub account, activate the SERP API trial, and copy your token from the API Playground. The Playground lets you pick the engine, keyword, country, and language, run the request, preview the structured response, and download it as JSON or HTML.
That live preview is underrated for evaluation. You can probe the verticals your team actually needs, compare full JSON against Light JSON sizes, and check geotargeted results before you commit any budget to the Thordata SERP API at all.
SDKs and Ready-Made Code Samples
Official SDKs exist for Python and Node.js, and the documentation carries paste-ready samples for Python, cURL, PHP, and Node.js. A basic call is a POST to scraperapi.thordata.com/request with a JSON body containing the target search URL, so even without an SDK the integration is an afternoon job rather than a sprint.
For teams wiring search collection into existing web scraping pipelines, the REST shape means the Thordata SERP API drops into Airflow jobs, cron workers, or queue consumers with no special client ceremony.
The Proxy-Mode Escape Hatch
Engineers who prefer to own the parsing can bypass the extraction layer entirely. A proxy-mode endpoint at scraping.thordata.com:30001 routes your own scripts through Thordata’s residential network, returning pages for your parsers instead of pre-extracted fields.
That split matters more than it looks. It means one contract covers both ends of the build-versus-buy spectrum: structured SERP data when you want it finished, raw network access when your stack already knows how to parse a results page itself.
Thordata SERP API Pricing: Only Success Is Billed
The pricing page lists six tiers, and the rule under them is the headline: only HTTP 200 responses are charged. Status codes 300, 400, 401, 403, 404, 429, 500 and 504 return nothing to your invoice, so a rough day against Google’s anti-bot systems cannot drain your balance for data you never received.
| Volume | Price / 1K responses | Monthly total |
|---|---|---|
| 500 (7-day trial) | $0.00 | Free, no card |
| 15,000 | $1.20 | $18 |
| 50,000 | $1.10 | $55 |
| 150,000 | $0.90 | $135 |
| 500,000 | $0.80 | $400 |
| 1,000,000 | $0.70 | $700 |
Unit cost falls almost forty percent between the smallest paid tier and the million-response tier. The chart below puts that volume curve in perspective.
What You Do Not Pay For
Errors are free by design. A 429 rate-limit response, a 500 upstream failure, or a 504 timeout is not billed, and neither are 403 blocks, which in a hostile scraping environment is the most common failure mode of all. Balance queries and status checks are free too, so monitoring your own consumption costs nothing.
The free trial adds 500 responses over seven days with no credit card, which is enough to validate every vertical your team cares about before the Thordata SERP API touches your budget.
Enterprise Beyond the Price List
Large or unusual workloads move to a custom Enterprise agreement: high concurrency, dedicated account managers, custom data fields, custom packages, and custom monitoring and alerting. Thordata’s published tiers cap at a million responses a month, so anything above that, or anything needing contractual SLAs, goes through the sales team.
That is a normal shape for the category. What is not normal, and worth repeating, is that even on the cheapest $18 plan the pay-on-success guarantee is identical to the enterprise terms, with no billing asterisks between tiers.
Thordata SERP API vs SerpApi, DataForSEO and Bright Data
The Category in One Table
The Thordata SERP API competes against three much older names: SerpApi, the San Francisco SERP specialist; DataForSEO, the data-infrastructure veteran with its own SERP API; and Bright Data, the largest proxy operator in the market, which also sells SERP scraping. The table summarises how they position in late 2026.
| Factor | Thordata SERP API | SerpApi | DataForSEO | Bright Data |
|---|---|---|---|---|
| Billing model | Success-only, flat tiers | Search quota plans | Pay-as-you-go tasks | Subscriptions, custom |
| Entry price | $18/mo (15K responses) | ~$75/mo (5K searches) | ~$1.70/1K tasks | Hundreds/mo, quote-led |
| Free trial | 500 responses, no card | 100 searches/mo | Small trial credit | Trial on request |
| Engines | Google, Bing, Yandex, DuckDuckGo | Google-centric, many engines | Google, Bing, Yahoo, more | Google, Bing, DuckDuckGo |
| AI-native tooling | MCP server, LangChain, n8n | Community wrappers | SDKs, no MCP | Agent and dataset tooling |
Where Each One Wins
Thordata wins on price-to-coverage at small and mid volumes: $18 a month for 15,000 responses undercuts every comparable service, and the success-only meter means that entry budget buys results, not retries. SerpApi wins on battle-tested reliability and ecosystem maturity, with over a decade of public reviews and the deepest Google coverage in the category.
DataForSEO wins when SERP data is one input among many, since one balance also funds domain analytics, keyword data, and content APIs. Bright Data wins where procurement demands a global compliance story and named enterprise support, and pays for it. Many teams pair an incumbent with the Thordata SERP API for cost-heavy long-tail ranks.
Testing Each Competitor’s Free Tier
SerpApi’s public free allowance is around 100 searches a month, enough for smoke tests but too thin for a pilot dataset. DataForSEO credits a small trial balance on sign-up, and Bright Data arranges evaluations through sales rather than self-serve. The Thordata SERP API trial, at 500 responses over seven days with no card, sits between the two extremes and is the only one sized for a real comparison week.
That asymmetry is worth exploiting. Run the same 500-query sample against the Thordata SERP API trial and against your incumbent’s free allowance, log block rates and payload quality, and you have a procurement-ready comparison before anyone buys anything from anyone.
Feeding AI Agents With the Thordata SERP API
The MCP Server Puts Search in the Agent
Thordata ships a Model Context Protocol server alongside the API, which means Claude Desktop, Cursor, and other MCP clients can call live search results as a native tool rather than through a hand-rolled integration. For agent builders, this is the cheapest way to give a model current web facts: no scraper code, no proxy config, just a tool definition.
That places the Thordata SERP API in a small early club. MCP support is becoming the connective tissue between AI assistants and data services, and a vendor that ships the server first gets pulled into agent workflows by default rather than by negotiation.
LangChain, LlamaIndex, n8n, Dify and the SDKs
Beyond MCP, Thordata publishes connectors for LangChain and LlamaIndex, so retrieval pipelines can treat the SERP API as a first-class loader. No-code shops reach the same data through n8n and Dify nodes, and the official Python and Node SDKs cover the custom-code path. Claude Desktop and the OpenAI SDK round out the integration list.
The practical effect is that a research agent can query Google News verticals, a price-monitoring agent can query Shopping verticals, and both share one meter. When websites increasingly decide which agents get through, as our piece on AI agents being blocked explains, structured APIs are the lane that stays open.
Why Agents Need Structured Search Anyway
Models hallucinate on current events, and browsing raw HTML pages burns context window on cookie banners and nav menus. A SERP API hands an agent ranked titles, snippets, and links in a few hundred compact tokens, which is exactly the shape of a grounding step. Light JSON exists for this reason, and agent builders are its newest audience.
As AI search reshapes what publishers see, with Google AI Search changing how publishing works, the demand side grows too: every rank tracker, every visibility monitor, and every agent harness needs someone to fetch and parse the page of results at scale.
Limitations, Compliance, and What to Watch
An English-Only Surface
Thordata’s website and documentation are English-only, with no localised equivalents today. For English-speaking engineering teams that is a non-issue, but it is worth flagging given the vendor’s China-adjacent base and the heavily multilingual markets its proxy network covers. Support quality is promised around the clock, though, with a dedicated technical team on enterprise accounts.
A Review Base Still Maturing
SerpApi and Bright Data carry thousands of third-party reviews; the Thordata SERP API has a smaller public footprint, with the parent proxy service more widely reviewed than the SERP product itself. The mitigations are concrete, since the trial requires no card and the pay-on-success rule caps downside on any given month, but procurement teams used to heavy review evidence should plan hands-on evaluation time.
Playing Within the Rules
Thordata’s own FAQ is direct that SERP scraping legality depends on terms of use and request behaviour, and advises staying inside search engine guidelines while avoiding excessive request frequency. You still own the compliance question your lawyer asks: what data you collect, how you store it, and what you do with it downstream is your side of the contract, whoever operates the proxies.
What to Watch in Your First Month
Instrument everything during evaluation: log the HTTP status of every response, the share of non-200s by vertical, and median latency by target country. Block rates and latency are the two numbers that decide whether a SERP provider is actually working, and they are easy to miss when the dashboard shows only volume spent.
Comparing those logs across providers is how agencies usually land their final choice. The Thordata SERP API will be judged on exactly those two curves, so capture them cleanly during the trial while the decision is still open and the data is still cheap.
Should You Use the Thordata SERP API?
Trial It This Week If…
You track ranks across cities or countries, you need Google verticals beyond organic results, or you are building agents that need live search without owning scraping infrastructure. At $18 a month for 15,000 responses, and with 500 free responses available before any card details are typed, the honest evaluation cost of the Thordata SERP API is roughly one hour of an engineer’s time.
Stay With an Incumbent If…
Your procurement process demands enterprise review evidence, you need SERP data as one slice of a broader marketing-data balance, or your contracts already standardise on SerpApi or Bright Data and nobody wants to reopen that discussion. Those are organisational reasons, not product reasons, and they are usually the right call at large-enterprise scale.
For everyone else in the middle, the sensible pattern is a one-month pilot: run your real workload, measure block rates against your own error logs, and let the success-only invoice make the business case for you.
A Sensible Pilot Plan
Start by pulling your current tracking list through the Thordata SERP API trial and diffing results against whatever you pay for today; mismatches tell you about localisation depth faster than any spec sheet. Then script one vertical you do not yet monitor, Maps or Shopping, and see whether it changes what your team reports.
Finally, wire the token into one non-critical job for a month, such as a weekly competitor rank export, and read the invoice. Because failed responses never appear on it, the pilot’s only real cost is the engineering hour you spend measuring.
FAQs About the Thordata SERP API
What does the Thordata SERP API do?
The Thordata SERP API fetches live search engine results pages from Google, Bing, Yandex and DuckDuckGo and returns them as structured JSON, Light JSON, or raw HTML. Proxies, CAPTCHA handling, rendering, and parsing all happen on Thordata’s side, so your code only sees clean data.
How much does the Thordata SERP API cost?
Paid tiers run from $18 a month for 15,000 responses ($1.20 per thousand) down to $700 a month for one million responses ($0.70 per thousand). A seven-day trial with 500 responses is free and needs no credit card, and only HTTP 200 responses are billed at any tier.
Which search engines and result types are supported?
The Thordata SERP API covers Google across seventeen verticals including Search, Maps, Shopping, Hotels, News, Trends, Scholar, Jobs, Lens and Flights. Bing adds Search, News, Shopping, Maps, Images and Videos, and Yandex and DuckDuckGo are supported as full engines through the same endpoint.
Can AI agents use the Thordata SERP API?
Yes. The Thordata SERP API ships with an MCP server for Claude Desktop, Cursor and other MCP clients, plus connectors for LangChain, LlamaIndex, n8n and Dify, official Python and Node.js SDKs, and a proxy-mode endpoint for agents that parse pages themselves.
References
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.
