What a SERP API actually costs per 1,000 searches
Credits, plan tiers and pagination make search API pricing hard to compare. Here is how to work out your real cost per 1,000 searches, with published list prices as of August 2026.
Search API pricing is quoted in credits, requests, searches and “results”, rarely in the one number you need: what you pay for a thousand searches at the volume you actually run.
Here is how to compute that number, and why the same vendor can look five times more expensive than a competitor at one volume and barely twice at another.
Step 1: convert everything into searches
Most APIs, including ours, price in credits, because different endpoints cost different amounts of work. A Google search is not the same job as fetching a static HTML page.
The conversion is one division:
searches per plan = credits included / credits per search
cost per 1,000 searches = plan price / searches per plan x 1000On FastAF a search costs 5 credits, so the Scale pack at $99 for 300,000 credits gives 60,000 searches, which is $1.65 per 1,000. The Pro pack at $299 for 1,000,000 credits gives 200,000 searches, which is $1.50 per 1,000. The cost per search drops as the pack grows, which is true of essentially every vendor in this market.
Step 2: compare at your volume, not at the headline
This is where most comparison tables mislead, including ones written about us.
SerpApi’s published plans, checked on their pricing page in August 2026, run from $75 for 5,000 searches up to $3,750 for 1,000,000. That is $15.00 per 1,000 at the Developer tier and $3.75 per 1,000 at the Cloud 1M tier: the same product, a 4x swing in unit price depending on which row you land on.
So the honest comparison depends entirely on your volume:
| Searches | FastAF | SerpApi | Ratio |
|---|---|---|---|
| 15,000 | $1.93 / 1k (Growth, $29) | $10.00 / 1k (Production, $150) | ~5x |
| 60,000 | $1.65 / 1k (Scale, $99) | $7.25 / 1k (Searcher, $725) | ~4x |
| 200,000 | $1.50 / 1k (Pro, $299) | $5.90 / 1k (Volume, $1,475) | ~4x |
| 1,000,000 | $1.40 / 1k (Business + 2x Pro, $1,397) | $3.75 / 1k (Cloud 1M, $3,750) | ~2.7x |
Each row uses the cheapest plan on either side that covers that volume, priced at full plan capacity. Note the rounding effect: if your volume falls just above a tier boundary you pay for the whole next plan, so the real unit cost at 60,001 searches is worse than the table suggests on both sides.
The gap is widest for small and mid-size projects and roughly halves at seven-figure volumes. Anyone quoting you a single “Nx cheaper” number across all volumes, us included, is quoting the row that flatters them.
Step 3: add the costs that are not on the pricing page
Three things routinely double a real bill after the plan is chosen.
Pagination. Google no longer returns 100 results in one request, so “track the top 100” costs ten requests per keyword, not one. Ten thousand keywords tracked weekly at depth 100 is 400,000 searches a month, not 40,000. Check whether your rank tracker’s “depth” setting is quietly multiplying your bill.
Failed requests. Ask whether a blocked or failed request is billed. On FastAF it is not: credits are reserved when the request starts and returned if it does not come back with a result. If a vendor bills attempts rather than results, your effective cost per usable record is higher than the sticker price by whatever their failure rate is, and their failure rate is not something you can audit.
Concurrency. Every plan carries a cap on parallel requests, and it is usually the constraint that decides how long a nightly job takes, not the credit quota. A plan with plenty of credits and 5 concurrent slots will not finish a 200,000-keyword sweep overnight. Compare the concurrency line as carefully as the price line.
Step 4: sanity-check against the floor
There is a price floor in this market. Running residential and mobile proxy pools, solving anti-bot challenges and maintaining parsers costs real money, and anyone selling far below the cost of the proxy traffic is either subsidising growth or serving you cached results. Cached search results are worse than no results: they look fine in a spreadsheet and silently corrupt rank tracking.
If a quote looks impossibly cheap, ask two questions: is this live at request time, and what is the success rate on hard targets.
The short version
Work out your monthly search volume first, convert every vendor’s plan into cost per 1,000 searches at that volume, then check pagination, failed-request billing and concurrency before signing anything. Our numbers are on the pricing page with the credit cost of every endpoint, and the SerpApi comparison shows the tradeoff at both ends of the volume range.