SerpApi Alternatives Priced in One Unit: A Thousand Google Searches

Every comparison of SERP APIs currently ranking for this query was written by a SERP API vendor. Scrapfly, Scrape.do, ScrapingDog, ScrapingBee, ScraperAPI, Crawlbase, Firecrawl, ScrapingAnt and half a dozen smaller shops all publish one, each concluding that the reader should buy from the publisher. That is not dishonest, and it does mean nobody neutral has priced the field.

The harder problem sits underneath the bias. These vendors do not sell the same unit, so the per-thousand figures they quote at each other are not comparable, and the conversion that makes them comparable is published by each vendor in a different place.

What follows prices four alternatives to SerpApi in one unit: a thousand Google search requests. Every figure comes from the named vendor's own pricing page, read on 30 August 2026, and every conversion between a vendor's unit and a Google search is stated so it can be checked.

The unit problem, in one example

ScraperAPI's cheapest plan is $49 a month for 100,000 API credits. Read as searches, that is 49 cents per thousand, which would make it fifty times cheaper than SerpApi's Starter tier.

It is not, and ScraperAPI's own FAQ says so: "A standard page costs 1 credit. Amazon costs 5, Google and Bing (including their subdomains) cost 25, and LinkedIn costs 30." Sites behind bot protection add 10 credits per request when bypassed.

One Google search on ScraperAPI therefore costs 25 credits. The $49 ScraperAPI plan buys 4,000 Google searches, not 100,000, which puts it at $12.25 per thousand. That single division changes the ranking of the whole field, and it is the kind of thing a vendor-authored comparison has no incentive to perform on itself.

The same discipline applies in both directions. Serper prices in credits where one credit is one query, so its own per-thousand figures need no conversion. Oxylabs prices its Web Scraper API per thousand results rather than per request, which is a third unit again.

The field, converted

The SerpApi and ScraperAPI rows are divisions performed here over each vendor's own price, allowance and credit schedule. The two Serper rows appear on Serper's page in that form already, under the banner "The World's Fastest & Cheapest Google Search API", so they are copied rather than calculated. Those five divisions are the only arithmetic in this article.

SerpApi StarterUSD 25.00 per thousand

• USD 25 a month for 1,000 searches
• One search costs one search

ScraperAPI HobbyUSD 12.25 per thousand

• USD 49 a month for 100,000 credits
• A Google request costs 25 credits, so 4,000 searches

Serper StarterUSD 1.00 per thousand

• USD 50 prepaid for 50,000 credits
• One credit is one query, and the rate is on Serper's own card

Serper, and the shape of a prepaid model

Serper publishes four prepaid credit bundles, under a line that reads "No monthly subscriptions; simply pay for the credits you need." Serper Starter is $50 for 50,000 credits at $1.00 per thousand. Serper Standard is $375 for 500,000 credits at $0.75 per thousand. Serper Scale is $1,250 for 2.5 million credits at $0.50 per thousand. Serper Ultimate is $3,750 for 12.5 million credits at $0.30 per thousand. Each tier has a rate limit, 50 queries per second on Starter rising through 100 and 200 to 300 on Ultimate, and each is marked "Credits valid for 6 months" with prices excluding sales tax, VAT and GST.

Two structural differences from a subscription matter more than the headline rate.

Credits do not reset monthly, they expire at six months. A team with lumpy demand can buy once and draw down across two quarters, which suits an occasional large list build far better than a monthly allowance that vanishes unused.

And the rate limit is decoupled from the balance. Serper's Starter tier allows 50 queries per second, which is 180,000 an hour, against a 50,000-credit bundle. That inverts the constraint found on subscription ladders where hourly throughput is pinned to a fraction of the monthly allowance, and it means a Serper bundle can be spent as fast as the caller can issue requests.

The trade is coverage. Serper's own banner calls it "The World's Fastest & Cheapest Google Search API", and its endpoints are Google surfaces: search, images, news, maps, places, videos, shopping, scholar, patents and autocomplete. A workflow that needs Amazon, an app store or Bing is not served by it, and that is precisely the breadth SerpApi's much higher rate is buying.

Serper also makes two claims worth separating from its prices: "In general, all queries are returned in 1 to 2 seconds." and that its prices are up to ten times cheaper than competitors including SerpApi and Bright Data. The first is a performance claim with no method attached. The second is a comparison the seller performs on itself, and the conversion table above suggests the multiple is larger than ten at the entry tier and depends entirely on which tiers are set against each other. A free allowance of 2,500 queries with no card required is the cheap way to test both claims rather than to accept them.

ScraperAPI, once the credits are converted

ScraperAPI's ladder runs Hobby at $49 a month for 100,000 credits with 20 concurrent threads, Startup at $149 for 1 million and 50 threads, Business at $299 for 3 million and 100 threads, Scaling at $475 for 5 million and 200 threads, Professional at $975 for 10.5 million and 300 threads, and Advanced at $1,975 for 21.5 million. Annual ScraperAPI billing takes ten percent off. A seven-day ScraperAPI trial with 5,000 credits requires no card.

At 25 credits per Google request the effective ScraperAPI rate for Google runs from $12.25 per thousand at the bottom to $2.30 at the top. That is genuinely cheaper than SerpApi at every comparable point, and it is a long way from the impression the credit headline creates.

Two things the credit model gives that a per-search plan does not. Geotargeting is a tier feature here rather than a parameter, restricted to the US and EU on the two cheapest plans and global from Business upward, which is a real constraint for anyone building lists outside those regions. And because a standard page costs one credit, the same balance covers ordinary web scraping alongside search, so a workflow that reads a search result and then fetches the pages behind it runs on one meter instead of two. That is the pattern in the infinite lead glitch, where a search API produces the URLs and a scraper reads the pages.

DataForSEO and Oxylabs, and what could not be read

DataForSEO prices pay as you go: "With DataForSEO you pay only for the individual services you consume." and, on the same page, "Note that the minimum payment amount is $50." There is no subscription and no plan ladder beside those two lines. Its per-endpoint prices live on separate pages that did not render figures to either a plain fetch or a headless browser render, so no DataForSEO per-search rate is quoted here. An unreadable page is a fact about the fetch rather than about the vendor, and the honest position is that the model is published and the numbers were not obtained.

Oxylabs presents a different problem, and it is on the vendor's own page rather than in the fetching. One Oxylabs pricing page gives the Web Scraper API as "Starts from $0.25 /1K results" in one section and "Starts from $49/mo" in another, and residential proxies at $2.5 per gigabyte in the first and $6 per gigabyte in the second. Both figures for each product are on the same document. Where a vendor's own page carries two numbers, the disagreement is the finding rather than something to resolve by picking the friendlier one, and a buyer should get the applicable rate in writing. Note also that a price per thousand results is not a price per thousand searches, since one search returns many results.

Making the numbers comparable

• ✓Yes: Establish what one unit is: a search, a query, a credit, or a result
• ✓Yes: Find the credit schedule, because a Google request can cost many credits
• ✓Yes: Check whether unused volume expires monthly or carries for months
• ✓Yes: Check whether throughput is pinned to the plan size or sold separately
• ~Depends: Check whether geotargeting is a parameter or a paid tier
• ~Depends: Check which engines beyond Google are actually covered

Choosing, without pretending there is one answer

Three shapes cover most go-to-market use, and they point at different providers.

Google only, at volume, on an occasional schedule. The prepaid model wins here on both rate and expiry, and Serper's own $1.00 per thousand at entry is the lowest committed entry rate in this set.

Search plus the pages behind it. A single credit balance that covers both is worth more than a marginally lower search rate, which favours a general scraping API over a pure SERP one.

Breadth, indemnity and certifications. SerpApi's own pricing is the expensive option in this set and it is buying things the others do not put on a card: coverage of more than a hundred endpoints across several engines, a legal indemnity on every plan card, and named security certifications. A team whose procurement asks who owns the risk is paying for a different product, not the same one at a worse rate.

What none of them changes is the quality of the list that comes out. A cheaper search API builds a badly targeted list faster and for less, which is the failure mode the GTM tools market map keeps circling: the tooling layer got cheap and the targeting did not get better on its own. The same caution applies to whatever orchestrates the calls, whether that is a script or a no-code extraction chain.

If the constraint is what happens after the list exists, get a free campaign plan rather than a cheaper API.

Pricing and features verified as of August 2026. Verify current terms with the vendor before relying on them.