Free SERP API Options in 2026: Free Tiers Compared, and Their Limits
Michael Carter · · 16 min read

Key takeaways
- On November 18, 2026, free SERP API tiers ranged from 20 Google searches (Scrapingdog) to 5,000 a month (Bright Data), after converting credits at each provider's own rate.
- Seven of the twelve SERP API providers we checked describe a one-time grant or trial; only Bright Data, Apify, SerpApi and Zenserp say their free allowance renews every month.
- Credits are not searches: a Google request costs 5 credits at Scrapingdog, 10 at ScrapingBee and 25 at ScraperAPI.
- Google's Custom Search JSON API still gives 100 free queries a day, but only to existing customers, who must move by January 1, 2027.
- One free BestSERP search on November 18, 2026 cost 1 credit and returned 9 US organic results, with every link a final URL 28 seconds after the request.
A free SERP API gives you a limited number of Google searches without paying. On November 18, 2026, free tiers ran from 20 searches (Scrapingdog) to 5,000 a month (Bright Data). Most are one-time sign-up grants meant for testing, only four renew monthly, and Google's own 100 free queries a day are closed to new customers.
This guide is for developers who want to test an idea or run a small job before paying: a dated table of 13 free tiers in each provider's own words, one real free search with its cost and timing, Google's own free options, SearXNG, and the catches that matter in production. We make BestSERP, so every other figure links to that provider's page. For paid plans, read our comparison of the best SERP API options.
Free SERP API options at a glance
The table converts each free allowance into Google results pages of up to 10 results, at the provider's own published rate per search. Each provider's name links to the page we read on November 18, 2026; "not stated" means that page does not say.
| Provider | Free allowance | Google searches | Renews | Card needed |
|---|---|---|---|---|
| Bright Data | 5K requests a month | 5,000 | Monthly | No |
| BestSERP (ours) | 3,000 searches, valid 30 days | 3,000 | Once | No |
| Serper | 2,500 queries | 2,500 | Once | No |
| Apify | $5 of usage a month | Up to 1,111 | Monthly | No |
| DataForSEO | $1 trial credit | 500 live, 1,666 queued | Once | Not stated |
| SerpApi | 250 searches a month | 250 | Monthly | Not stated |
| ScraperAPI | 5,000 credits, 7-day trial | 200 | Once | No |
| ScrapingBee | 1,000 credits | 100 | Once | No |
| SearchApi | 100 requests | 100 | Once | No |
| Zenserp | 50 searches a month | 50 | Monthly | Not stated |
| Scrapingdog | 100 credits | 20 | Not stated | No |
| Oxylabs | $5 welcome credits | Not stated | Once | Not stated |
| Google Custom Search JSON API | 100 queries a day | 100 a day | Daily | Existing customers only |

Three things stand out:
- Most free tiers are a one-time grant. Seven of the twelve providers give credits once, at sign-up or as a trial. Only Bright Data, Apify, SerpApi and Zenserp say their allowance comes every month.
- Credits are not searches. A Google request costs 5 credits at Scrapingdog, 10 at ScrapingBee and 25 at ScraperAPI, so 1,000 credits can mean 40 searches.
- The largest monthly tier is Bright Data's. Its card says "5K requests" in the heading and "5K records per month" below, with no other limit stated.
Free search APIs that do not return Google results
These search their own index, not Google's page, which suits many AI and research jobs (our Tavily alternatives comparison covers them):
| API | Free allowance | Searches | Card needed |
|---|---|---|---|
| Brave Search API | $5 of credits every month | 1,000 at $5 per 1,000 | Yes: "To prevent abuse." |
| Tavily | 1,000 credits every month | 1,000 basic searches | No |
| Exa | $10 of credits every month | "up to 2,500 Lite searches" | No |
What "free" means: monthly tiers, sign-up grants and trials
A free tier is an allowance of requests you can spend before paying, and its shape matters more than its size: a monthly tier can run a small job indefinitely, while a one-time grant only tests an integration.
- Monthly allowances reset each cycle. When Apify's $5 runs out, "your access to Apify's services will be blocked until the beginning of the next monthly cycle" (Apify pricing).
- Sign-up grants arrive once. BestSERP's 3,000 free searches are "Valid for 30 days after sign-up" (pricing); Serper's page does not say when its 2,500 free queries expire.
- Trials are grants on a short clock: ScraperAPI's "7-day trial and 5,000 API credits", then a free plan with 1,000 credits.
- Dollar credits depend on the mode: DataForSEO's $1 buys 500 live results pages at $0.002 or 1,666 queued ones at $0.0006 (its Google Organic pricing).
How long each free tier lasts at a steady volume, computed from the table above:
| Provider | At 100 searches a day | At 1,000 a day |
|---|---|---|
| Bright Data | The whole month | 5 days a month |
| BestSERP (ours) | 30 days, once | 3 days, once |
| Serper | 25 days, once | 2.5 days, once |
| Apify | 11 days a month | 1 day a month |
| DataForSEO (live) | 5 days, once | 12 hours, once |
| SerpApi | 2.5 days a month | 6 hours a month |
| ScraperAPI | 2 days, once | 5 hours, once |
| ScrapingBee, SearchApi | 1 day, once | 2 hours, once |
| Zenserp | 12 hours a month | 1 hour a month |
| Scrapingdog | 5 hours, once | 29 minutes, once |
At 1,000 searches a day, no free SERP API tier lasts a full week, so plan the paid step from the start. Our SERP API pricing comparison prices that step at every volume.
What one free search buys: a real, tested example
One free search buys one Google results page: up to 10 organic results with titles, links, snippets and site names, plus the video carousel and Top stories when the page has them. On November 18, 2026 at 04:35 UTC we spent 1 free BestSERP credit on "free google search api" for the US, through the adapter shown later (Python 3.12.10, requests 2.34.2):
query 'free google search api', gl=us, hl=en, 2026-11-18 04:35 UTC
cost: 1 credit(s), cached: False, search ran from: us
answer: 9 organic results, 0 videos, 0 top stories
time: durationMs 4,968, 28 s until every link was final (linksPending False, unresolvedLinks 0)
balance after: 2,087 credits; next expiry 2026-12-17
1 developers.googleblog.com Check out the Custom Search API
2 olostep.com Best Web Search APIs: Compare Top Search API Tools
3 stackoverflow.com Is there a Google Search API I can call, server-side, no ...
4 faun.dev Top 23 Google Custom Search JSON API Alternatives & ...
5 neilsk.medium.com Build Your Own Google Search Engine in Under 30 Minutes

What the run shows:
- Cost. 1 credit for one page, not from cache;
num: 20would cost 2 (how credits work). - Speed. The search took 5.0 seconds (
durationMs). The adapter then polledGET /requests/{requestId}(free) untillinksPendingwas false: all nine links were final URLs 28 seconds after the request. - Country.
ipCountrywasus: the search ran from a US address, as asked. - Scale. At 1 credit a page, 3,000 free credits are 3,000 pages like this one: 100 keywords checked daily for the 30 days they last, enough to try a daily rank tracker in Python.
To see what is left and when it expires, call GET /account, which costs nothing (tested the same day):
import os
from datetime import datetime, timezone
import requests
resp = requests.get(
"https://api.bestserp.com/account",
headers={"X-API-KEY": os.environ["BESTSERP_API_KEY"]},
timeout=130,
)
resp.raise_for_status()
account = resp.json()
now = datetime.now(timezone.utc)
print(f"Balance: {account['balance']:,} credits, limit {account['rateLimitPerMinute']} searches a minute")
for grant in account["grants"]:
expires = grant["expiresAt"]
if expires is None:
when = "never expires"
else:
days = (datetime.fromisoformat(expires) - now).days
when = f"expires {expires[:10]} ({days} days left)"
print(f"- {grant['source']}: {grant['remaining']:,} of {grant['amount']:,} left, {when}")
Balance: 2,076 credits, limit 60 searches a minute
- signup: 2,076 of 3,000 left, expires 2026-12-17 (29 days left)
The account is shared with our other tests, which is why the balance fell between the two runs.
Google's own free quota, and why it is closing
Google's Custom Search JSON API still "provides 100 search queries per day for free", but only for existing customers: Google's Custom Search JSON API overview says it "is closed to new customers" and that existing customers "have until January 1, 2027 to transition to an alternative solution." The pricing ($5 per 1,000 extra queries, up to 10,000 a day) applies only to them until then.
That is why the search above is a useful warning. Six of the nine results for "free google search api" are about this API, led by a Google blog post from June 12, 2008, and only one of their six snippets says it is closed to new customers. Google's January 20, 2026 announcement points JSON API users to Vertex AI Search "for up to 50 domains" and asks anyone who needs full web search to contact Google.
If you are an existing customer, our Custom Search JSON API migration guide maps its fields to a SERP API, and the state of official Google search APIs covers every other Google option.
You may not need an API: free options from Google and from us
Many jobs that seem to need a free SERP API are covered by a free tool:
| Your need | Free option | The limit |
|---|---|---|
| How your own site ranks | Google Search Console | Your verified sites only; average positions |
| A search box on your website | Programmable Search Element, "Sites to search" | 50 domains at most |
| Answers with citations in an LLM app | Gemini API Grounding with Google Search | 5,000 searches a month; display-only terms |
| A few rank checks in another country | BestSERP SERP Checker | 3 checks a day, 5 keywords, top 20 |
- Search Console. Search Console is free, and its Search Analytics API allows 1,200 queries a minute per site (usage limits). It reports your own average positions, not other sites' results (why they differ from a rank tracker's).
- Programmable Search Element. Google's January 2026 announcement calls "Sites to search" a "free feature" for up to 50 domains; whole-web engines must move by January 1, 2027. It is a widget for visitors, not a data feed.
- Gemini grounding. Grounding with Google Search "includes 5,000 free search requests per month shared across all Gemini 3 and newer models", then $14 per 1,000. It returns a model answer with citations, not a result list, and the Gemini API terms (effective March 23, 2026) allow showing grounded results only to the user who asked and forbid caching or analyzing them.
- Checking by hand. Our free SERP Checker shows where a domain ranks in the top 20 for up to 5 keywords in any country; checking rankings in another country compares the free methods.
Self-hosted and open-source options: what SearXNG really returns
SearXNG is the usual free, self-hosted answer, and it is a metasearch engine, not a SERP API. The SearXNG documentation calls it "a free internet metasearch engine which aggregates results from up to 260 search services"; it runs from one Docker command on your own server.
What to know before you rely on it:
- JSON is off by default. The search API returns JSON with
format=jsononly whenjsonis listed undersearch.formatsinsettings.yml(the default lists onlyhtml). "Requesting an unset format will return a 403 Forbidden error," and "many public instances have these formats disabled." - Google sees your server. Its Google engine queries Google from your server's IP address, so you take on the blocking a SERP API handles for you.
- Blocked engines are switched off for hours or days. The default settings suspend an engine for 86,400 seconds (1 day) after a CAPTCHA, 3,600 seconds after "Too many request" or HTTP 429, and 604,800 seconds (7 days) after a Google CAPTCHA.
- The list is a blend. Because it aggregates several engines, the result order is SearXNG's, not Google's positions, so it does not suit rank tracking.
We did not run SearXNG for this post: its Google engine would send searches to Google from our own IP, which our test rules do not allow. The same blocking problem breaks free Python libraries that scrape Google directly; why googlesearch-python fails with 429 errors shows it, and SERP API vs web scraping compares the full cost of running your own scraper.
Catches to check before you build on a free SERP API
Free tiers differ more in their rules than in their size. These eight checks come from the providers' own pages, read on November 18, 2026:
| Catch | What we found | What to do |
|---|---|---|
| Expiry | BestSERP: 30 days. ScraperAPI: 7-day trial. Serper: not stated | Test in the first week |
| Credits per search | Scrapingdog 5, ScrapingBee 10 (light request), ScraperAPI 25 | Convert credits to searches first |
| Throughput | SerpApi Free: 50 searches an hour. Scrapingdog: 1 concurrent request. BestSERP: 60 a minute, same as paid | Match your burst size |
| At zero | Serper "will stop accepting new queries"; Apify blocks until the next cycle; BestSERP answers 402 insufficient_credits |
Handle the error in code |
| Card | Brave asks for one; most others say no card is needed | Use a card limit if asked |
| Free-plan gaps | SerpApi's Free card excludes ZeroTrace Mode and priority support; Zenserp lists Bing, Yandex and DuckDuckGo on paid plans only | Test the plan you will buy |
| What is charged | SerpApi does not count cached or failed searches; BestSERP charges a cached answer like a fresh one | Cache repeats yourself |
| Use of results | Gemini grounding: display only; Brave: storing results "with a custom plan" | Read the terms for your use |
Can a free tier run in production? A monthly allowance can carry a small, steady job that fits inside it, such as 250 SerpApi searches or Bright Data's 5,000 requests a month. A one-time grant cannot: it proves an integration works before you pay.
Moving from a free SERP API to paid without rewriting code
Keep the provider behind one function, so moving from one free tier to another, or to a paid plan, is a configuration change. Your app calls search() and gets the same five fields whichever API answered.
This adapter ran against BestSERP on November 18, 2026 (Python 3.12.10, requests 2.34.2) and produced the example above. The SerpApi adapter follows SerpApi's Google Search API docs and was not run: we have no SerpApi account.
import os
import time
from urllib.parse import urlparse
import requests
LAST = {} # the provider's last raw answer, for logging
def bestserp(q, gl, hl):
# Serper-style request: POST /search, JSON body, X-API-KEY header, camelCase fields.
api, headers = "https://api.bestserp.com", {"X-API-KEY": os.environ["BESTSERP_API_KEY"]}
for _ in range(3):
r = requests.post(f"{api}/search", headers=headers, json={"q": q, "gl": gl, "hl": hl}, timeout=130)
while r.status_code == 202: # still running: polling is free
time.sleep(2)
r = requests.get(f"{api}/requests/{r.json()['requestId']}", headers=headers, timeout=130)
if r.status_code in (429, 502, 503): # rate limited, failed or busy (502/503 are refunded)
time.sleep(int(r.headers.get("Retry-After", 5)))
continue
if r.status_code == 402: # the free (or paid) credits are used up
raise RuntimeError("out of credits: top up, or set SERP_PROVIDER to another adapter")
r.raise_for_status()
data = r.json()
for _ in range(24): # final URLs arrive by polling until linksPending is false (free, up to 2 min)
if not data.get("linksPending"):
break
time.sleep(5)
poll = requests.get(f"{api}/requests/{data['requestId']}", headers=headers, timeout=130)
if poll.ok:
data = poll.json()
LAST["bestserp"] = data
return [{"position": o["position"], "title": o["title"], "url": o["link"],
"site": o.get("domain"), "snippet": o.get("snippet", "")}
for o in data.get("organic", [])]
raise RuntimeError(f"search failed after 3 tries: HTTP {r.status_code}")
def serpapi(q, gl, hl):
# Not run for this post: written from SerpApi's docs (engine, q, gl, hl, api_key; organic_results).
r = requests.get("https://serpapi.com/search", timeout=130, params={
"engine": "google", "q": q, "gl": gl, "hl": hl, "api_key": os.environ["SERPAPI_API_KEY"]})
r.raise_for_status()
LAST["serpapi"] = r.json()
return [{"position": o["position"], "title": o["title"], "url": o["link"],
"site": urlparse(o["link"]).hostname, "snippet": o.get("snippet", "")}
for o in r.json().get("organic_results", [])]
ADAPTERS = {"bestserp": bestserp, "serpapi": serpapi}
def search(q, gl="us", hl="en"):
"""The only function your app calls. Returns [{position, title, url, site, snippet}]."""
return ADAPTERS[os.environ.get("SERP_PROVIDER", "bestserp")](q, gl, hl)
Three details make the switch easy:
- Handle 402 and 429 on purpose. The end of a free tier is an expected error, not an outage.
- Normalize field names. SerpApi's list is
organic_results; BestSERP's and Serper's isorganic. - Serper-format clients move with two values. BestSERP accepts Serper's paths, body and field names, so a Serper client changes its base URL and key (migration notes); we did not test this against Serper itself.
How we compared the free SERP API tiers
On November 18, 2026, between about 04:15 and 04:45 UTC, we read each provider's own pricing page, and the docs page that sets credits per Google search, with a web fetcher. Because several pages draw their tables in the browser, we also rendered every page in Chromium and saved its visible text. Every quote above is copied from those pages.
- Unit: one Google web results page of up to 10 results at the provider's default options (ScrapingBee's light request, Scrapingdog's Google SERP Light API, DataForSEO's live mode unless stated).
- Who: the ten providers of our best SERP API comparison, plus ScraperAPI and Zenserp, which the top-ranking US article for "free SERP API" lists, and Google's own API. Brave, Tavily and Exa are listed apart because they do not return Google results.
- Left out: promotions and programs you apply for (startup or student credits), Brave's banner offer of "Up to $20 in free calls every month" (its Search plan card says $5), annual discounts and taxes.
- Not tested: we did not open accounts with the other providers, so their success rates, speed and enforcement of limits are untested. The one search above is our own product.
Free tiers change without notice, so check the linked page before you sign up.
Next steps
Pick the free SERP API tier whose shape fits the job: a monthly allowance for a small recurring task, a sign-up grant to test an integration, and no API at all when Search Console or a free checker answers the question.
- Paid options, coverage and verdicts: the best SERP API options compared and SERP API cost at every scale.
- Our own rates: 3,000 free searches, then $0.50 per 1,000 with credits that never expire, on the BestSERP pricing page.